ABSTRACT
The development of tissues and organs requires close interaction of cells. To achieve this, cells express adhesion proteins such as the neural cell adhesion molecule (NCAM) or its Drosophila ortholog Fasciclin 2 (Fas2). Both are members of the Ig-domain superfamily of proteins that mediate homophilic adhesion. These proteins are expressed as isoforms differing in their membrane anchorage and their cytoplasmic domains. To study the function of single isoforms, we have conducted a comprehensive genetic analysis of Fas2. We reveal the expression pattern of all major Fas2 isoforms, two of which are GPI anchored. The remaining five isoforms carry transmembrane domains with variable cytoplasmic tails. We generated Fas2 mutants expressing only single isoforms. In contrast to the null mutation, which causes embryonic lethality, these mutants are viable, indicating redundancy among the different isoforms. Cell type-specific rescue experiments showed that glial-secreted Fas2 can rescue the Fas2 mutant phenotype to viability. This demonstrates that cytoplasmic Fas2 domains have no apparent essential functions and indicate that Fas2 has function(s) other than homophilic adhesion. In conclusion, our data suggest novel mechanistic aspects of a long-studied adhesion protein.
INTRODUCTION
The development of a complex organism requires manifold concerted interactions of the cellular building blocks. These interactions in part depend on regulated adhesion, which is of particular importance in the nervous system (Silies and Klämbt, 2011). Here, neurons form intricate networks in which uncountable specific cellular connections provide the hardware of neuronal computation (Eichler et al., 2017; Larderet et al., 2017; Schneider-Mizell et al., 2016). Full functional complexity of the brain, however, then requires glial cells, which interact with each other and with neurons to perform a multitude of different tasks required for neuronal function (Yildirim et al., 2019). The close cell-cell interactions observed in the nervous system are based on two properties – contact and signaling. Cell-cell contacts and adhesion need to be specified to ensure the formation of stable connections and subsequently allow cell-cell signaling and communication. Cell communication is exemplified by chemical synapses, where neurons exchange information by secreting and receiving transmitter molecules. In addition, neurons can communicate in manifold ways with glial cells (Gundersen et al., 2015). In most cases, neuron-glia interaction depends on close contact of the respective cells. Therefore, adhesion proteins are of particular relevance to allow the interaction between different cell types.
Both homophilic and heterophilic adhesions can mediate cellular interactions. An example of the latter type is Amalgam, a secreted adhesion protein with three tandem immunoglobulin (Ig) domains. In solution, Amalgam forms dimers that cross-link Neurotactin proteins expressed on the surfaces of opposing cells (Frémion et al., 2000; Liebl et al., 2003; Zeev-Ben-Mordehai et al., 2009). Similarly, heterophilic interaction has been reported for Neurexin IV/Caspr which – depending on an alternatively spliced exon – can either bind the Ig-domain protein Wrapper or Contactin, another Ig-domain protein (Noordermeer et al., 1998; Stork et al., 2009; Wheeler et al., 2009). Homophilic interactions are, for example, mediated by the Drosophila Ig-domain protein Neuroglian (Nrg), which constitutes the Drosophila homolog of the vertebrate L1 Ig-domain adhesion protein. Interestingly, different Nrg/L1 isoforms are characterized by distinct cytoplasmic domains and have been evolutionarily conserved. These two adhesion proteins (Nrg167 and Nrg180 in the fly) are expressed by either glial or neuronal cells and in both cell types provide the link to the actin cytoskeleton via common adaptor proteins (Bieber et al., 1989; Chen and Hing, 2008; Hortsch et al., 1990; Yamamoto et al., 2006). Another well-studied homophilic adhesion protein known in vertebrates and invertebrates is neural cell adhesion molecule (NCAM) or Fasciclin 2 (Fas2). Fas2 was originally discovered as a motor axon marker that was subsequently shown to mediate activity-dependent expansion of the neuromuscular junction (Davis et al., 1997; Schuster et al., 1996a,b; Thomas et al., 1997).
Fas2/NCAM are evolutionarily very well conserved. Interestingly, Ncam mouse mutants are viable and fertile with only minor nervous system phenotypes (Cremer et al., 1994). In contrast, Drosophila Fas2 mutants are early larval lethal and display some morphological phenotypes at the developing neuromuscular junction (Kohsaka et al., 2007; Kristiansen and Hortsch, 2010; Schuster et al., 1996a,b). It has been suggested that Fas2 allows neuronal plasticity to influence circadian behavior (Sivachenko et al., 2013). Importantly, both, Fas2 and its vertebrate homolog NCAM are involved in neuron-glia signaling (Higgins et al., 2002; Paratcha et al., 2003; Wright and Copenhaver, 2001). In Drosophila, a graded expression of Fas2 on motor axons is needed for peripheral glial migration (Silies and Klämbt, 2010). Although gain of Fas2 expression clearly halts glial cell migration, glial migration appears normal in Fas2 null mutants. Outside of the nervous system, Fas2 is needed for microvilli length and organization in the Malpighian tubules, stabilizing the brush border and possibly requiring homophilic adhesion (Halberg et al., 2016). The Fas2 gene is predicted to encode at least seven differentially expressed isoforms. As known for NCAM, the different Fas2 isoforms differ mostly in their membrane attachment and the cytoplasmic domain. However, despite the wealth of genomic information on the different Fas2 isoforms, their possible individual contribution is not known in any context.
Here, we have conducted a comprehensive genetic analysis of the different Fas2 isoforms to decipher their relative contributions during development. Our data reveals that only the GPI-anchored Fas2PB affects glial migration, but otherwise highly redundant functions of all isoforms were identified. Interestingly, both neuronal expression of transmembrane-anchored Fas2PD as well as glial expression of secreted Fas2 rescues the lethal Fas2 phenotype, suggesting that Fas2 can act independently of its homophilic adhesion functions.
RESULTS
Distinct Fasciclin 2 isoforms are GPI anchored
The Fasciclin 2 (Fas2) gene spreads over 75 kbp and encodes a series of evolutionarily conserved Ig-domain proteins (Fig. 1A-C). Currently, seven mRNA splice variants encoding seven distinct proteins have been described (Fas2PA, Fas2PB, Fas2PC, Fas2PD, Fas2PF, Fas2PG and Fas2PH; Fig. 1A,B; FlyBase). To verify the presence of the predicted splice variants, we extracted total mRNA from wild-type embryos and performed RT-PCR reactions with isoform-specific primer pairs (Fig. 1D). cDNA corresponding to all seven isoforms could be amplified with isoforms RB, RC and RF being predominant (Fig. 1D).
Different isoforms of the Fasciclin 2 locus. (A) Schematic view of the Fas2 locus organization. Transcription is from right to left. Only one promoter region is present. The positions of several exon trap or MiMIC transposon insertions that have been utilized to determine the expression pattern of the different Fas2 isoforms are depicted. The exons are color-coded corresponding to the seven Fas2 isoforms. (B) Higher magnification of the 3′ end of the Fas2 gene. The organization of the different isoforms is indicated. Isoforms PB and PC are linked by a GPI anchor. All other isoforms share a common transmembrane domain and variable cytoplasmic domains. (C) Organization of the principle Fas2 proteins. FN, fibronectin type III domain; Ig, Ig domain. The color coding is as shown in A,B. The vertical blue lines in the protein structure indicate the position of the introns. (D) Expression of all predicted isoforms was tested by RT-PCR using exon-specific primer combinations. cDNA fragments of the expected length were detected that differed from amplification products using genomic DNA. RA/RD (genomic: 1500 bp, cDNA: 300/390 bp); RA (genomic: 850 bp, cDNA: 240 bp); RB (genomic 1500 bp, cDNA: 560 bp); RC (genomic 680 bp, cDNA: 580 bp); RF (genomic: 11,154 bp, cDNA: 600 bp); RG (genomic: 750 bp, cDNA 240 bp); RH (genomic: 500 bp, cDNA 220 bp). Expression of the isoforms PB, PC and PF can be detected most robustly. (E-L) Staining of stage 15 embryos (n>10 per genotype) carrying different exon trap elements that allow detection of the GFP-tagged Fas2 isoforms as indicated. Embryos are stained for expression of GFP to detect expression of the endogenously tagged Fas2GFP fusion proteins (green), Repo to label glial nuclei (red) and HRP (blue) to show all neuronal membranes. Stars in F,H denote epidermal Fas2 expression, arrowheads indicate glial staining. Scale bar: 20 µm.
Different isoforms of the Fasciclin 2 locus. (A) Schematic view of the Fas2 locus organization. Transcription is from right to left. Only one promoter region is present. The positions of several exon trap or MiMIC transposon insertions that have been utilized to determine the expression pattern of the different Fas2 isoforms are depicted. The exons are color-coded corresponding to the seven Fas2 isoforms. (B) Higher magnification of the 3′ end of the Fas2 gene. The organization of the different isoforms is indicated. Isoforms PB and PC are linked by a GPI anchor. All other isoforms share a common transmembrane domain and variable cytoplasmic domains. (C) Organization of the principle Fas2 proteins. FN, fibronectin type III domain; Ig, Ig domain. The color coding is as shown in A,B. The vertical blue lines in the protein structure indicate the position of the introns. (D) Expression of all predicted isoforms was tested by RT-PCR using exon-specific primer combinations. cDNA fragments of the expected length were detected that differed from amplification products using genomic DNA. RA/RD (genomic: 1500 bp, cDNA: 300/390 bp); RA (genomic: 850 bp, cDNA: 240 bp); RB (genomic 1500 bp, cDNA: 560 bp); RC (genomic 680 bp, cDNA: 580 bp); RF (genomic: 11,154 bp, cDNA: 600 bp); RG (genomic: 750 bp, cDNA 240 bp); RH (genomic: 500 bp, cDNA 220 bp). Expression of the isoforms PB, PC and PF can be detected most robustly. (E-L) Staining of stage 15 embryos (n>10 per genotype) carrying different exon trap elements that allow detection of the GFP-tagged Fas2 isoforms as indicated. Embryos are stained for expression of GFP to detect expression of the endogenously tagged Fas2GFP fusion proteins (green), Repo to label glial nuclei (red) and HRP (blue) to show all neuronal membranes. Stars in F,H denote epidermal Fas2 expression, arrowheads indicate glial staining. Scale bar: 20 µm.
The different Fas2 isoforms are predicted to be either transmembrane proteins or tethered to the membrane via glycosylphosphatidylinositol (GPI) anchors but real experimental evidence for the latter is missing. Isoforms PA, PD, PG, PH and PF all share one common exon that is absent from PB and PC (Fig. 1A,B). Of the former isoforms, PA is the same as the PD isoform but in addition carries a cytoplasmic PEST domain. This PEST domain is also present in isoform PG, which contains an additional 12 amino acids in the cytoplasmic domain. The PH isoform carries a unique cytoplasmic domain of 27 amino acids, and isoform PF is generated by adding 15 amino acids to the first Ig domain of the PD isoform (Fig. 1A,B).
Isoform PC is predicted to have a transmembrane anchor (SMART analysis, smart.embl-heidelberg.de); however, it had been suggested that Fas2PC is linked to the plasma membrane by a GPI anchor (Grenningloh et al., 1991). Isoform PB lacks a clear transmembrane domain and a GPI anchor is predicted by a Kohonen self-organizing map, GPI-SOM (Fankhauser and Mäser, 2005).
To determine experimentally the membrane anchorage of Fas2PB and Fas2PC, we generated hemagglutinin (HA)-tagged cDNA clones and expressed the proteins in S2 cells. HA-Fas2PB and HA-Fas2PC were found in cell lysates at the expected size of 90 kDa in western blots (Fig. S1). Minor amounts of both Fas2 protein isoforms are released from the cell membrane and can be detected in the supernatant. To test whether the proteins are linked to the plasma membrane via a GPI anchor, we added phosphatidylinositol-specific phospholipase C (PiPLC) to the cell culture medium. This enzyme, which specifically cleaves GPI anchors (Heinz et al., 1998), was able to efficiently solubilize HA-Fas2PB as well as HA-Fas2PC protein, which was then found in the supernatant (Fig. S1). This demonstrates that the Fas2 isoforms PB and PC are attached to the plasma membrane by a GPI anchor. The exon common to all other isoforms (PA, PD, PF, PG and PH) contains a transmembrane domain and, therefore, we will refer to these isoforms together as Fas2TM (Fig. 1C).
Fas2 isoforms are differentially expressed
In order to detect Fas2 protein expression within the developing nervous system, we utilized a number of protein-trap insertion lines and one MiMIC strain, which we used to insert a monomeric Cherry (mCherry)-encoding exon (MiMIC12989; Nagarkar-Jaiswal et al., 2015) (Figs 1A and 2). The insertion of an mCherry-encoding exon in MiMIC12989 (Fas2MiMIC12989::mCherry) reduced fitness but a homozygous stock could be established.
Expression of all Fas2 isoforms as detected by MiMIC12989-Cherry. Expression of a Fas2::mCherry fusion generated by inserting a Cherry exon into the MiMIC12989 element (n>10 animals were analyzed). (A-H) Expression of Fas2::mCherry in stage 16 embryos. (A,B) Lateral view. Anterior is to the left, dorsal is up. HRP staining (blue) detects all neuronal membranes, Repo (green) marks all glial nuclei. During embryonic development, robust Cherry expression can be detected outside of the ventral nerve cord (vnc) in the epidermis (epi), tracheal cells (tr) and the midgut (mg). Scale bar: 50 µm. (C,D) Ventral view. The boxed area is shown at higher magnification in E,F. (E,F) Fas2::mCherry is expressed by glial cells. This can be seen at the segmental nerves (arrows) and the glial blood-brain barrier (arrowheads). The dashed line indicates the position of the orthogonal sections shown in G,H. Scale bar: 50 µm. (I-L) Expression of Fas2::mCherry in the third instar larval brain. (I,J) Broad expression is found in the larval brain. Note that the characteristic expression pattern of Fas2 in a subset of axonal fascicles is not visible in the focal plane shown. Instead, expression in the ensheathing glia, which encase the entire neuropil, is visible (arrowhead, eg). The boxed area is shown at higher magnification in K,L. (K,L) Expression of Fas2::mCherry in glial cells. Expression can be seen in glial cells of the blood-brain barrier (arrowheads). It cannot be resolved whether perineurial (pg) and subperineurial glial cells (spg) express Fas2::mCherry. (M,N) Expression in the eye-imaginal disc. Neuronal and glial expression can be detected. Glial expression is found in the perineurial glia (pg, arrowheads) and the wrapping glia (arrows). (I-N) Scale bars: 20 µm.
Expression of all Fas2 isoforms as detected by MiMIC12989-Cherry. Expression of a Fas2::mCherry fusion generated by inserting a Cherry exon into the MiMIC12989 element (n>10 animals were analyzed). (A-H) Expression of Fas2::mCherry in stage 16 embryos. (A,B) Lateral view. Anterior is to the left, dorsal is up. HRP staining (blue) detects all neuronal membranes, Repo (green) marks all glial nuclei. During embryonic development, robust Cherry expression can be detected outside of the ventral nerve cord (vnc) in the epidermis (epi), tracheal cells (tr) and the midgut (mg). Scale bar: 50 µm. (C,D) Ventral view. The boxed area is shown at higher magnification in E,F. (E,F) Fas2::mCherry is expressed by glial cells. This can be seen at the segmental nerves (arrows) and the glial blood-brain barrier (arrowheads). The dashed line indicates the position of the orthogonal sections shown in G,H. Scale bar: 50 µm. (I-L) Expression of Fas2::mCherry in the third instar larval brain. (I,J) Broad expression is found in the larval brain. Note that the characteristic expression pattern of Fas2 in a subset of axonal fascicles is not visible in the focal plane shown. Instead, expression in the ensheathing glia, which encase the entire neuropil, is visible (arrowhead, eg). The boxed area is shown at higher magnification in K,L. (K,L) Expression of Fas2::mCherry in glial cells. Expression can be seen in glial cells of the blood-brain barrier (arrowheads). It cannot be resolved whether perineurial (pg) and subperineurial glial cells (spg) express Fas2::mCherry. (M,N) Expression in the eye-imaginal disc. Neuronal and glial expression can be detected. Glial expression is found in the perineurial glia (pg, arrowheads) and the wrapping glia (arrows). (I-N) Scale bars: 20 µm.
Three Fas2 insertion lines are available that, in principle, should detect the expression of all isoforms: Fas2GFP778, Fas2CPTI000483 (Lowe et al., 2014) and Fas2MiMIC12989::mCherry. During embryogenesis, the protein-trap insertion CPTI000483 directed robust expression in a subset of neurons. In addition, weak expression was detected in ectodermal cells (Fig. 1E,F). A similarly broad GFP expression pattern was associated with the protein-trap insertion Fas2GFP778, in the embryo as well as in the larval nervous system (Fig. S2). In Fas2MiMIC12989::mCherry animals, broader Cherry expression could be detected, suggesting that the position of the additional GFP exon present in CPTI000483 and Fas2GFP778 affects splicing probability in the different cell types, with Fas2MiMIC12989::mCherry providing a better read-out of Fas2 expression (Fig. 2). In Fas2MiMIC12989::mCherry embryos, expression could be noted in a subset of neurons but was also seen in the epidermis, the Malpighian tubules, the midgut and in glial cells (Fig. 2A-H). This exon trap thus reveals that Fas2 is more broadly expressed than previously known.
The exon-trap insertion line Fas2GFP397 labeled all expression domains except that of Fas2RC. It revealed strong GFP expression in the nervous system, including expression in glial and epidermal cells (Fig. 1G,H; Silies and Klämbt, 2010). The insertion CPTI001279 (Lowe et al., 2014) resulted in GFP labeling of isoforms PA, PD, PG and PH, and showed an exclusively neuronal expression pattern, which also corresponded to the expression pattern detected by the monoclonal antibody mAb 1D4 (Grenningloh et al., 1991) (Fig. 1I,J). The difference between these two expression patterns can therefore be attributed to isoform PB, arguing that PB is expressed in glial cells, and other tissues.
A similar expression as observed in CPTI001279 was detected by the insertion CB03613, which labels some but not all transmembrane isoforms. This insertion disrupts the PEST domain-encoding exon and showed an only neuronal expression (Fig. 1K,L). Very similar expression profiles were detected during larval stages (Fig. S2). The two insertion lines Fas2GFP778 and Fas2CPTI000483, which, in principle, should detect expression of all isoforms were mostly expressed in neurons with some epidermal expression. No clear glial expression was seen (Fig. S2A-H). In contrast, Fas2MiMIC12989::mCherry also showed clear neuronal and glial Fas2 expression. In the third instar brain, surface glial cells expressed Fas2MiMIC12898::mCherry, but the broad expression pattern did not allow discrimination between perineurial and subperineurial glial cells. In addition, the ensheathing glia that surround the neuropil expressed Fas2MiMIC12898::mCherry (Fig. 2I-L). In the eye-imaginal disc, the strongest Fas2MiMIC12898::mCherry expression was noted in photoreceptor cells. Expression could also be detected in glial cells, which could be identified as perineurial glial cells based on their position at the base of the eye disc (Fig. 2M,N). The insertion Fas2GFP397 showed clear ectodermal expression in leg and eye-antennal imaginal discs and also revealed glial expression (Fig. S2I-L). As noted in the embryo, CPTI001279 and CB03613 directed expression exclusively in neuronal cells (Fig. S2M-T). In summary, Fas2 is more broadly expressed than previously thought, and Fas2MiMIC12898::mCherry presents a useful tool to study the full Fas2 expression pattern. Whereas the transmembrane-anchored Fas2 proteins PA and PD-PH appear to have a purely neuronal expression profile, the GPI-anchored isoform PB also shows additional glial and epidermal expression. In addition, isoform PC is also found in the tracheal system (see below). Importantly, all differences in Fas2 isoform expression are due to differential splicing in the different cell types.
Generation of a molecularly and genetically defined Fas2 null mutant
A previously generated, transposon excision-induced Fas2 allele, Fas2EB112, carries a 1.7 kbp deletion in the presumed Fas2 promoter region. Fas2EB112 homozygous mutant animals die as early first instar larvae and expression of Fas2TM, which can be detected by the monoclonal antibody 1D4, is absent (Grenningloh et al., 1991). Given that this antibody does not detect the expression of other isoforms, and because it was unclear whether the promoter deletion would also affect the expression of all other isoforms, we generated a Fas2 null allele that was clearly defined at the molecular level. We used a recombination-based approach (Parks et al., 2004) employing FRT elements residing in P{XP}d07035, inserted 300 bp upstream of the presumed transcriptional start site, and PBac{WH}f06654, inserted downstream of the Fas2 locus (Fig. 2A). Thereby a small deletion covering the entire Fas2 gene locus plus one additional gene, GlcAT-I, was generated (Fig. 3A). GlcAT-I encodes a glycosyltransferase that is ubiquitously expressed throughout development (Kim et al., 2003). The PBac insertion in the 5′ UTR of GlcAT-I (PBac{WH}GlcAT-I [f00247]) is viable and fertile in homozygosis. In contrast, hemizygous Df(1)Fas2 animals are early first instar larval lethal with no detectable Fas2 protein expression. The identical lethal phases and the same glial migration phenotype (see below) noted for both Fas2 alleles indicates that the classic Fas2EB112 is indeed a Fas2 null allele, and the observed phenotypes are due to loss of Fas2 function.
Generation of Fasciclin 2 null alleles and rescue experiments. (A) Schematic view of the Fas2 locus organization as described in Fig. 1. The triangles indicate the insertion of the different transposons used during the mutant analysis. The turquoise triangles represent FRT-bearing transposons used to generate deficiencies Df(1)Fas2 and Df(1)Fas2TM as indicated. All Bac-based duplications are shown. (B-K) Immunohistochemical analysis of embryos (B-G) and larval brain with eye-imaginal discs (H-K). Scale bars: 20 µm. n>10 animals analyzed. Anti-Fas2 (mAb 1D4), and anti-Repo are shown in magenta (or gray), GFP (B,C) and anti-HRP (D,F,H,J) staining is in green. (B,C) In heterozygous stage 14 (B), or stage 16 (C) embryos of the genotype Df(1)Fas2TM/FM7twiGFP Fas2 expression can be detected. Weak GFP expression due to the balancer is detected. (D-G) In contrast, in Df(1)Fas2TM mutant embryos (D, stage 14; F, stage 16), no expression of the anti-1D4 antigen is detected. Loss of membrane-bound Fas2 proteins does not lead to abnormal glial or neuronal phenotypes. (H-K) Similarly, in Df(1)Fas2TM mutant larval brains (H,I), or eye-imaginal discs (J,K) no abnormal neural phenotypes can be detected. Note that no Fas2 staining can be detected because the cytoplasmic Fas2 domain is deleted and thus the mAb 1D4 antigen is absent.
Generation of Fasciclin 2 null alleles and rescue experiments. (A) Schematic view of the Fas2 locus organization as described in Fig. 1. The triangles indicate the insertion of the different transposons used during the mutant analysis. The turquoise triangles represent FRT-bearing transposons used to generate deficiencies Df(1)Fas2 and Df(1)Fas2TM as indicated. All Bac-based duplications are shown. (B-K) Immunohistochemical analysis of embryos (B-G) and larval brain with eye-imaginal discs (H-K). Scale bars: 20 µm. n>10 animals analyzed. Anti-Fas2 (mAb 1D4), and anti-Repo are shown in magenta (or gray), GFP (B,C) and anti-HRP (D,F,H,J) staining is in green. (B,C) In heterozygous stage 14 (B), or stage 16 (C) embryos of the genotype Df(1)Fas2TM/FM7twiGFP Fas2 expression can be detected. Weak GFP expression due to the balancer is detected. (D-G) In contrast, in Df(1)Fas2TM mutant embryos (D, stage 14; F, stage 16), no expression of the anti-1D4 antigen is detected. Loss of membrane-bound Fas2 proteins does not lead to abnormal glial or neuronal phenotypes. (H-K) Similarly, in Df(1)Fas2TM mutant larval brains (H,I), or eye-imaginal discs (J,K) no abnormal neural phenotypes can be detected. Note that no Fas2 staining can be detected because the cytoplasmic Fas2 domain is deleted and thus the mAb 1D4 antigen is absent.
Fas2 can act without its cytoplasmic domain
We next asked if deletion of all isoforms or of specific Fas2 isoforms was required for the lethality caused by loss of Fas2. We used molecularly defined chromosomal duplications to test the genetic requirements of Fas2. Bacterial artificial chromosome (BAC)-based duplications are stretches of genomic DNA with an average length of 88 kb. Many of these constructs are inserted in the same landing site on chromosome 3L, allowing their direct comparison (Venken et al., 2010). Four duplications were tested, Dp(1;3)DC115, Dp(1;3)DC465, Dp(1;3)DC075 and Dp(1,3)Fas2, which we generated by integrating the BAC clone CH321-75A18 into the landing site VK33 (Fig. 3A). Except Dp(1;3)DC075, all duplications cover the entire Fas2 gene with all exons but differ in the length of their 5′ sequences. Dp(1;3)DC075 lacks the exons located on the 3′ end of the Fas2 gene resulting in a construct that only expresses the Fas2PC isoform (Fig. 3A).
Except Dp(1;3)DC075, all duplications fully rescue the lethality associated with Fas2EB112. Interestingly, Dp(1;3)DC115 harbors only short 5′ sequences upstream of the predicted promoter suggesting that most regulatory sequences reside in intronic regions. The presence of Dp(1;3)DC075 alone still provided a partial rescue with 10% of the expected number of hemizygous flies appearing [17 Fas2EB112/Y; Dp(1:3)DC075 males and 164 FM7/Y; Dp(1:3)DC075 males]. This argues that a GPI-anchored isoform alone can fulfill important Fas2 functions. To test this more explicitly, we used genome editing to generate a CRISPR-induced mutation leaving only isoform Fas2PC intact (Fas2ΔPB, ΔTM) (see Materials and Methods). This mutant was fully viable with no obvious discernible phenotypes at this level of resolution (Fig. 4, Table 1). The rescue ability of the Dp(1;3)DC075 duplication might be not as effective because the mRNA may be truncated owing to the nature of the construct and is possibly not as stable as the mRNA generated from the CRISPR-induced allele. In conclusion, however, both experiments suggest that expression of only the GPI-anchored isoform Fas2PC is sufficient to restore viability.
Generation of isoform-specific Fasciclin 2 mutants. (A) Schematic view of the Fas2 locus with the position of all CRISPR-induced mutations indicated by red arrows. Some of the mutants were generated in the background of the exon trap line Fas2GFP397, which allows the detection of all Fas2 isoforms except Fas2PC. (B,C) Third instar larval brain of a homozygous Fas2PC mutant stained for Fas2TM (anti-1D4, magenta/gray) and Repo (green). No abnormal neural phenotype can be detected. (D,E) Third instar larval brain of a homozygous Fas2PB mutant stained for Fas2TM (anti-1D4, magenta/gray) and Repo (green). No abnormal neural phenotype can be detected. Scale bar: 20 µm.
Generation of isoform-specific Fasciclin 2 mutants. (A) Schematic view of the Fas2 locus with the position of all CRISPR-induced mutations indicated by red arrows. Some of the mutants were generated in the background of the exon trap line Fas2GFP397, which allows the detection of all Fas2 isoforms except Fas2PC. (B,C) Third instar larval brain of a homozygous Fas2PC mutant stained for Fas2TM (anti-1D4, magenta/gray) and Repo (green). No abnormal neural phenotype can be detected. (D,E) Third instar larval brain of a homozygous Fas2PB mutant stained for Fas2TM (anti-1D4, magenta/gray) and Repo (green). No abnormal neural phenotype can be detected. Scale bar: 20 µm.
Fas2 can act independently of its homophilic adhesion properties
In addition, we employed an isoform-specific overexpression approach for rescue experiments (see Materials and Methods for details). Several transgenes encoding different Fas2-YFP proteins have been published (Kohsaka et al., 2007). Expression of wild-type, YFP-tagged transmembrane Fas2PD protein in neurons and muscle cells has been shown to rescue the lethality associated with the Fas2EB112 mutant (Kohsaka et al., 2007).
In a next step, we expressed YFP-tagged Fas2PD specifically in glial cells using the repo-Gal4 driver. Glial expression of the transmembrane Fas2 transgene did not rescue the lethality of Fas2 mutants. In a cross of [Fas2EB112/FM7; UAS-Fas2PD-YFP×repo-Gal4] we found 63 FM7 males and no Fas2EB112 males. However, surprisingly, expression of a secreted Fas2 protein lacking all membrane anchorage sequences (Kohsaka et al., 2007) only in glial cells was able to rescue some animals to fully viable and fertile males. From the cross [Fas2EB112/FM7; UAS-Fas2extra-YFP×repo-Gal4] we found 167 FM7 males and ten fertile males (genotype: Fas2EB112/Y, UAS-Fas2extra-YFP; repo-Gal4). Likewise, neuronal expression of Fas2extra-YFP alone was able to only weakly rescue the lethal Fas2 phenotype. From the cross [Fas2EB112/FM7; UAS-Fas2extra-YFP×nsyb-Gal4] we found 143 FM7 males and five fertile males of the genotype Fas2EB112; UAS-Fas2extra-YFP; nsyb-Gal4. Because expression of a secreted Fas2 protein only in glial cells can also rescue lethality, we conclude that Fas2 does not strictly require a homophilic adhesive interaction to perform its essential role during development.
Genome editing generates isoform-specific Fas2 mutants
The P-element Fas2CB03613 (Buszczak et al., 2007) was inserted within an exon incorporated in Fas2PA and Fas2PG. Thus, the insertion disrupts at least the function of Fas2PA and Fas2PG. Hemizygous as well as homozygous flies were viable. To further delineate the requirement of the different Fas2 isoforms we first generated a deficiency chromosome in which only the GPI-anchored Fas2PB and Fas2PC proteins were present [Df(1)Fas2ΔTM; Fig. 3A], again using FRT-mediated recombination (Parks et al., 2004). Whereas expression of the isoform Fas2TM-specific mAb 1D4 epitope was visible in heterozygous Df(1)Fas2ΔTM embryos (Fig. 3B,C), it was absent from all homozygous tissues tested (Fig. 3D-K), confirming the loss of all isoforms carrying the large cytoplasmic domain. Df(1)Fas2ΔTM flies producing only GPI-anchored Fas2 proteins were viable and fertile in homozygosis with a reduced fitness compared with the balancer-carrying heterozygous animals. This argues that the transmembrane-tethered Fas2 isoforms and their cytoplasmic domains are not strictly required. Together, these data suggest that no signaling function originates from the cytoplasmic Fas2 domain to ensure viability.
In a next step, we utilized CRISPR/Cas9-based genome editing to induce further isoform-specific mutants by generating small deletions in the respective exons (see Materials and Methods). In all cases, the design of the sgRNA was aimed to disrupt the membrane anchorage sequences of the different Fas2 isoforms: either the predicted GPI anchor site of PB or PC, or the transmembrane domain common to all other isoforms (Fig. 4A). A mutation in the exon that is specific for the isoform Fas2PC caused a frameshift with a stop codon at amino acid position 742. Because this mutation removes the predicted hydrophobic domain needed for GPI anchorage it is expected to result in a secreted Fas2PC protein (Fas2ΔPC; see Materials and Methods for details for all mutants generated). These mutants were homozygous viable and fertile and expression of Fas2TM as recognized by the 1D4 mAb was normal at this level of resolution (Fig. 4A-C, Table 1). A mutation of the exon encoding a Fas2PB-specific GPI-anchor attachment site is also predicted to result in a secreted Fas2PB isoform. Hemizygous or homozygous flies lacking GPI-anchored Fas2PB were fully viable (Fas2ΔPB; Fig. 4A,D,E, Table 1). However, we noted a change in the expression of the 1D4 antigen (which is only present in Fas2TM) and mutant embryos displayed a glial migration phenotype during embryonic development (see below).
Fasciclin 2 expression in isoform-specific Fas2 mutants
We next sought to test how the loss of specific isoforms affects the expression of other isoforms. To visualize the expression of the Fas2 isoforms PB, PA, PD-PH in the mutant backgrounds that we generated, we utilized the Fas2GFP397 exon-trap strain, which confers broad expression in glial and neuronal cells as can be seen best on the segmental nerves in the embryo (Fig. 5A-C). We then generated a CRISPR/Cas9-mediated small deletion in the exon encoding the transmembrane domain in the Fas2GFP397 background (ΔTM; see Figs 1 and 4A), blocking the membrane anchorage of PA, PD-PH and thereby generating a predicted secreted Fas2 protein. In the absence of the transmembrane-anchored proteins Fas2TM, we noted robust expression of Fas2GFP along glial cells accompanying the peripheral nerves (Fig. 5D-F). To test how glial Fas2 expression interacts with axonal Fas2 expression, we next generated a CRISPR/Cas9 mutant specifically affecting the GPI anchorage of Fas2 in the Fas2397 background. In the absence of GPI anchorage of the Fas2PB protein, glial expression was gone, but neuronal expression of the Fas2GFP protein was still present (Fig. 5G-I). Similar Fas2GFP397 expression patterns could be detected during larval stages (Fig. S3A-E). Control Fas2GFP397 eye-imaginal discs showed both a glial and a neuronal expression domain (Fig. 5J,K). Upon genetic ablation of the GPI-anchored Fas2PB form, neuronal expression of Fas2 remained (Fig. 5L-O), whereas ablation of the neuronally expressed Fas2 proteins revealed the glial expression domain (Fig. 5P,Q). In the third instar larval nervous system, expression was more diffuse and could not be easily assigned to glial or neuronal cell types (Fig. S3F-J). The GFP signal may also originate from the Fas2TM::GFP isoforms now being secreted into the extracellular space. Consistent with this, in the eye-imaginal disc, strong expression of secreted Fas2, trapped between peripodial membrane and disc epithelium, could be detected (Fig. S3H,K-M). Together, these data further support neuronal expression of transmembrane Fas2 and glial expression of GPI-anchored Fas2PB. Moreover, expression in one cell type can be maintained without Fas2 expression in the other cell type.
Fas2PB is expressed by glial cells. (A-C) Homozygous Fas2GFP397 embryo stained for Fas2::GFP expression (green), Repo (red) and HRP (blue). The dashed area is shown at higher magnification in B,C. Note that little GFP expression is found on glial cell bodies (arrows). Most GFP expression is found along segmental nerves (arrowheads). (D-F) Homozygous Fas2GFP397, Fas2ΔTM mutant embryo. The dashed area is shown at higher magnification in E,F. Note the broader expression of GFP, which is not confined to segmental nerves, but is found in the entire glial cell body. (G-I) Homozygous Fas2GFP397, Fas2ΔPB mutant embryo. Neuronal expression of Fas2 is still visible. Note the even expression of Fas2 along the motor axon. Glial expression is absent. (J,K) Third instar larval Fas2GFP397 eye-imaginal disc stained for expression of GFP (J) and HRP (K). (L-O) Third instar larval Fas2GFP397, Fas2ΔPB mutant eye-imaginal discs. (L,N) GFP expression is found along photoreceptor axons. (M,O) HRP expression (M) and 1D4 expression (O) label photoreceptor neurons. (P,Q) Third instar larval Fas2GFP397, Fas2ΔTM mutant eye-imaginal disc. (P) Note the broad GFP expression in glial cells (arrowhead). Asterisk denotes expression in photoreceptor neurons. (Q) Neuronal HRP expression is normal. Scale bars: 20 µm. n>10 embryos (A-I) or eye imaginal discs (J-Q) were analyzed for each genotype.
Fas2PB is expressed by glial cells. (A-C) Homozygous Fas2GFP397 embryo stained for Fas2::GFP expression (green), Repo (red) and HRP (blue). The dashed area is shown at higher magnification in B,C. Note that little GFP expression is found on glial cell bodies (arrows). Most GFP expression is found along segmental nerves (arrowheads). (D-F) Homozygous Fas2GFP397, Fas2ΔTM mutant embryo. The dashed area is shown at higher magnification in E,F. Note the broader expression of GFP, which is not confined to segmental nerves, but is found in the entire glial cell body. (G-I) Homozygous Fas2GFP397, Fas2ΔPB mutant embryo. Neuronal expression of Fas2 is still visible. Note the even expression of Fas2 along the motor axon. Glial expression is absent. (J,K) Third instar larval Fas2GFP397 eye-imaginal disc stained for expression of GFP (J) and HRP (K). (L-O) Third instar larval Fas2GFP397, Fas2ΔPB mutant eye-imaginal discs. (L,N) GFP expression is found along photoreceptor axons. (M,O) HRP expression (M) and 1D4 expression (O) label photoreceptor neurons. (P,Q) Third instar larval Fas2GFP397, Fas2ΔTM mutant eye-imaginal disc. (P) Note the broad GFP expression in glial cells (arrowhead). Asterisk denotes expression in photoreceptor neurons. (Q) Neuronal HRP expression is normal. Scale bars: 20 µm. n>10 embryos (A-I) or eye imaginal discs (J-Q) were analyzed for each genotype.
Fas2PC is expressed in trachea, which are defective in Fas2 mutant embryos
To determine the expression pattern of Fas2PC, we inserted a short stretch of genomic DNA harboring a Fas2PC exon tagged with V5 and a Fas2PB exon tagged with HA into the Fas2MiMIC12989 insertion (Fig. 6A). Using these transgenic reporter lines, we noted Fas2PC expression in tracheal cells (Fig. 6B-E) and a small subset of neuronal cells (Fig. 6F-I). During larval stages, weak expression in glial cells could be detected in the eye-imaginal discs. In addition, expression was noted in the morphogenetic furrow of the eye-imaginal disc, in young photoreceptor neurons and in a crescent of larval brain cells that might correspond to the forming lamina (Fig. 6J-L). The embryonic expression pattern of Fas2PC prompted us to re-analyze the lethal phase of Fas2 null mutants. Fas2EB112 hemizygous embryos develop normally to stage 16 and move extensively in the egg shell suggesting that neuromuscular junctions are established and functional. Whereas control animals hatch after 24 h development, mutant Fas2 larvae move in the egg shells but never manage to hatch. Even after 48 h the mutant animals are still moving but cannot hatch. We found that the trachea is not inflated with air in Fas2EB112 mutants, which likely causes hypoxia (Fig. 7).
Fas2PC is expressed in non-neuronal cells. (A) Schematic view of the Fas2 locus, showing the construct used to follow Fas2PC expression inserted into the Mi{MIC}Fas2MI12989 piggybac insertion. (B-E) Stage 15 embryos expressing V5-tagged Fas2PC. Anterior is to the left. (B,C) Lateral view, dorsal up; (D,E) ventral view. Scale bar: 50 µm. Note the strong expression of Fas2PC in cells of the tracheal system (arrow) and some neurons. (F,G) High magnification of a stage 14 embryo showing Fas2PC expression in the CNS. Scale bar: 20 µm. (H,I) High magnification of a stage 16 embryo showing Fas2PC expression in the CNS. (J-L) Expression of Fas2PC in a third instar larval brain (J) and the attached eye-antennal discs (K,L). Note the strong expression in a crescent of cells in the larval brain lobes (arrowheads), the morphogenetic furrow (mf) and expression in glial cells of the eye-imaginal disc (arrows in K,L). Scale bar: 50 µm, n>10 embryos (A-I) or eye imaginal discs (J-Q) were analyzed for each genotype.
Fas2PC is expressed in non-neuronal cells. (A) Schematic view of the Fas2 locus, showing the construct used to follow Fas2PC expression inserted into the Mi{MIC}Fas2MI12989 piggybac insertion. (B-E) Stage 15 embryos expressing V5-tagged Fas2PC. Anterior is to the left. (B,C) Lateral view, dorsal up; (D,E) ventral view. Scale bar: 50 µm. Note the strong expression of Fas2PC in cells of the tracheal system (arrow) and some neurons. (F,G) High magnification of a stage 14 embryo showing Fas2PC expression in the CNS. Scale bar: 20 µm. (H,I) High magnification of a stage 16 embryo showing Fas2PC expression in the CNS. (J-L) Expression of Fas2PC in a third instar larval brain (J) and the attached eye-antennal discs (K,L). Note the strong expression in a crescent of cells in the larval brain lobes (arrowheads), the morphogenetic furrow (mf) and expression in glial cells of the eye-imaginal disc (arrows in K,L). Scale bar: 50 µm, n>10 embryos (A-I) or eye imaginal discs (J-Q) were analyzed for each genotype.
Fas2 mutants have a defective tracheal system. (A) Heterozygous Fas2EB112 embryo at 18 h old. The tracheal system including the dorsal trunk (arrowheads) is formed but not yet inflated. (B) Heterozygous Fas2EB112 embryo at 23 h old, shortly before hatching. The tracheal system is now filled with air. (C) Hemizygous Fas2EB112 mutant embryo at 48 h old. The tracheal system (arrowheads) is still not filled with air. Scale bar: 50 µm. n>10 embryos were analyzed for each genotype.
Fas2 mutants have a defective tracheal system. (A) Heterozygous Fas2EB112 embryo at 18 h old. The tracheal system including the dorsal trunk (arrowheads) is formed but not yet inflated. (B) Heterozygous Fas2EB112 embryo at 23 h old, shortly before hatching. The tracheal system is now filled with air. (C) Hemizygous Fas2EB112 mutant embryo at 48 h old. The tracheal system (arrowheads) is still not filled with air. Scale bar: 50 µm. n>10 embryos were analyzed for each genotype.
GPI-anchored Fas2 isoforms are needed for glial migration
Although neuronal expression is broadly intact when glial expression is lost, interaction between neuronal and glial Fas2 might still be needed for coordinated interaction between the two cell types. A hallmark of Fas2 expression is the graded expression of Fas2TM along motor axons (Fig. 8A-C,M). This gradient of neuronal Fas2 expression was previously shown to be required for the migration of peripheral glial cells during embryogenesis (Silies and Klämbt, 2010). To quantify the graded axonal distribution of Fas2, we determined the Fas2 expression in hemisegments 2-6 and then calculated the mean of Fas2 intensity ratio per embryo, which was used for subsequent statistical analysis. In brief, at the onset of glial migration during stage 14, 1.79 times more Fas2 was incorporated at the growing tip of the axon compared with the axonal membrane at the CNS/PNS transition zone (Fig. 8M). Interestingly, deletion of the Fas2PB isoform was sufficient to level the graded distribution of Fas2TM along motor axons (Fas2ΔPB, Fas2 expression intensity ratio of 1.26 along the axon; Fig. 8D-F,M). Given the above finding that Fas2PB is a glial-expressed protein, this implies that glial cells are able to affect the graded expression of a cell adhesion protein along motor axons. To further test this hypothesis, we followed the expression of Fas2TM in Df(2L)200 mutants; this mutation removes both glial cells missing 1 and glial cells missing 2 and thus all glial cells are absent (Hosoya et al., 1995; Jones et al., 1995; Kammerer and Giangrande, 2001; Vincent et al., 1996). In these embryos, the gradient of neuronal Fas2TM expression was also strongly affected (Fas2 expression intensity ratio of 1.03 along the axon; Fig. 8G-I,M), suggesting that a trans-interaction between glial cells and neurons is needed to set up the graded distribution of Fas2TM along the axons.
Graded expression of the neuronal Fas2TM isoform is caused by glial Fas2 expression. (A-L) Expression of the Fas2TM epitope (anti-1D4 staining) in stage 14 embryos (magenta). Neuronal membranes are in green (A,G) or glial cells are in green (D,J). False-color images represent the expression strength of Fas2 along the motor axons (C,F,I,L). (A-C) In wild-type embryos, Fas2TM is expressed in a graded fashion along the axon, with a stronger expression towards the growth cone. (D-F) This gradient is lost in homozygous Fas2ΔPB mutant embryos. (G-I) Likewise, no graded Fas2TM expression can be detected in a gcm, gcm2 mutant background. (J-L) The graded distribution of Fas2TM is only slightly affected by removing the expression of Fas2PC. (M) Statistical analysis of the Fas2 intensity ratio on motor neurons at the CNS/PNS boundary compared with expression in the growth cone area (Mann–Whitney Rank Sum test). Box-whisker plot represents the median (bold line), 25% and 75% quartiles (box), and 5% and 95% extreme values (circles). n, number of embryos tested per genotype; the number given in brackets represents the number of analyzed hemisegments. Scale bar: 20 µm. 8-20 embryos with 40-88 segments were analyzed as indicated.
Graded expression of the neuronal Fas2TM isoform is caused by glial Fas2 expression. (A-L) Expression of the Fas2TM epitope (anti-1D4 staining) in stage 14 embryos (magenta). Neuronal membranes are in green (A,G) or glial cells are in green (D,J). False-color images represent the expression strength of Fas2 along the motor axons (C,F,I,L). (A-C) In wild-type embryos, Fas2TM is expressed in a graded fashion along the axon, with a stronger expression towards the growth cone. (D-F) This gradient is lost in homozygous Fas2ΔPB mutant embryos. (G-I) Likewise, no graded Fas2TM expression can be detected in a gcm, gcm2 mutant background. (J-L) The graded distribution of Fas2TM is only slightly affected by removing the expression of Fas2PC. (M) Statistical analysis of the Fas2 intensity ratio on motor neurons at the CNS/PNS boundary compared with expression in the growth cone area (Mann–Whitney Rank Sum test). Box-whisker plot represents the median (bold line), 25% and 75% quartiles (box), and 5% and 95% extreme values (circles). n, number of embryos tested per genotype; the number given in brackets represents the number of analyzed hemisegments. Scale bar: 20 µm. 8-20 embryos with 40-88 segments were analyzed as indicated.
We then tested whether loss of any of the two GPI-anchored isoforms is required for a proper distribution of axonal Fas2 and tested if the isoform Fas2PC also affects the distribution of the Fas2TM isoform. In Fas2ΔPC mutants, only a slight change in the establishment of the Fas2TM gradient along motor axons could be detected (Fas2 intensity ratio 1.49; Fig. 8J-M). Thus, the two GPI-anchored isoforms Fas2PB and Fas2PC differentially contribute to the stabilization of the Fas2TM gradient on motor axons.
We next asked if the alteration in the Fas2TM distribution in the absence of glial Fas2 resulted in glial migration phenotypes. In wild-type embryos, peripheral glial nuclei steadily migrate towards the periphery (see Movie 1). In all mutants in which the expression of the Fas2PB isoform was affected, final positioning of glial nuclei appeared normal again at the end of migration in stage 16 embryos (Fig. S4). To test if there are more subtle phenotypes at the onset of migration, we performed in vivo live imaging of glial migration of stage 14-16 embryos. Interestingly, in mutants affecting the graded expression of Fas2TM, we noted a wiggling of glial nuclei during the short period of cell migration towards the periphery (see Table 1, and compare tracks in Fig. S4A,B, Movies 2-6). Glial cells moved backwards towards the CNS for a short period of time before resuming their outwards migration (see white and orange arrowheads in Fig. S4C-H). We did not note such a phenotype in mutants affecting Fas2PC expression. Together, these data argue that the correct distribution of Fas2, and especially glial Fas2 expression, is required for a coordinated and directed onset of migration. At the same time, mechanisms are in place that allow these initial phenotypes to be overcome during subsequent development.
DISCUSSION
The Drosophila NCAM homolog Fasciclin 2 (Fas2) was first identified as a motor neuron-specific protein (Grenningloh et al., 1991). Previously, we showed that Fas2 has a broader expression profile and is found in glial and renal cells but no clear analysis of the different Fas2 isoforms was carried out (Halberg et al., 2016; Silies and Klämbt, 2010). The newly generated Fas2MiMIC12898::mCherry allele, which labels all Fas2 isoforms, shows an even broader expression pattern, including the tracheal system. Further phenotypic studies revealed a previously unknown tracheal phenotype of Fas2 mutants. In addition, we analyzed isoforms Fas2PB and Fas2PC and show that they are both tethered to the membrane via a GPI anchor. We reveal distinct expression profiles of the different isoforms, and particularly show that Fas2PB is expressed in glial cells. Our data demonstrated functional redundancy of the different isoforms. Finally, rescue experiments demonstrate that Fas2 can act independently of its homophilic adhesion properties.
In the CNS, the GPI-linked isoforms Fas2PB and Fas2PC are found in non-neuronal cells (glia and trachea) and neurons, whereas the transmembrane-linked isoforms are generated exclusively in neurons. A similar expression profile has been noted in mammals. Here, the GPI-linked NCAM form, NCAM120, is expressed predominantly by glial cells whereas the neuronal forms, NCAM180 and NCAM140, are expressed by neurons (Maness and Schachner, 2007). In contrast to mutations in the mammalian NCAM gene (Cremer et al., 1994), mutations in Drosophila Fas2 result in a lethal phenotype (Grenningloh et al., 1991). Here, we show that the lethality associated with Drosophila Fas2 mutants might not be due to severe motor dysfunction. Instead, our data suggest that it might be due to disrupted tracheal development as we noted defective tracheal inflation in Fas2EB112 mutants. Similar phenotypes are caused by mutations that affect the very large extracellular protein Uninflated (Zhang and Ward, 2009) and by mutations affecting the chitin deacetylases Serpentine or Vermiform (Luschnig et al., 2006; Wang et al., 2006). Importantly, here we have only addressed two phenotypic traits – lethality and glial migration – but have not studied other functions of Fas2 during the development of the neuromuscular junction or the brush border in Malpighian tubules (Halberg et al., 2016; Kohsaka et al., 2007; Thomas et al., 1997).
Expression of the Fas2 gene is controlled by one promoter, suggesting that the differential expression during development is brought about by differential splicing. Such a cell type-specific difference in splicing activity has also been demonstrated for tramtrack and Neurexin IV. Some proteins controlling nervous system-specific splicing, such as HOW and Crooked Neck, have been identified that may also control the neural specific splicing of additional RNAs (Edenfeld et al., 2006; Giesen et al., 1997; Rodrigues et al., 2012).
Deletion mutants affecting expression of all Fas2 isoforms result in late embryonic or early larval lethality. A small deletion mutant that removed five of the seven isoforms, all of which are transmembrane proteins, is still viable with no discernible phenotype. The same is true for animals generated by reintroducing a truncated Fas2 locus allowing the expression of only one of the seven isoforms, the GPI-linked Fas2PC protein. A drawback in these mutant and rescue experiments is that we do not know how the splicing pattern is affected by the different mutants. Do glial cells that normally activate Fas2PB expression now also activate Fas2PC? We do not favor this idea because in mutants affecting Fas2PB expression we note a subtle glial migration phenotype, suggesting that no compensatory Fas2 expression by an altered cell type-specific splicing is induced.
Thus, one might anticipate that the different Fas2 isoforms act in a redundant manner to mediate adhesion. Such an adhesive interaction could even be mediated by a secreted Fas2 protein that, after binding to the cell surface via another interactor, might still be able to mediate adhesive cell-cell interactions. Alternatively, Fas2 might be a signaling molecule and could act non-cell-autonomously. The results of our cell type-specific rescue experiments support this model. Whereas expression of membrane-anchored Fas2 in glial cells does not rescue lethality associated with the Fas2 null phenotype, glial expression of secreted Fas2 is able to rescue the lethal Fas2 phenotype. Based on these findings, we postulate that Fas2 has some signaling function perceived by an as-yet-unknown receptor to allow survival of the animal.
In hippocampal neurons from mice it has been shown that NCAM not only mediates homophilic cell-cell interactions but can also trigger exocytosis via the FGF receptor (Chernyshova et al., 2011). In addition, NCAM is able to bind and hydrolyze extracellular adenosine triphosphate, which also exerts important signaling functions (Dzhandzhugazyan and Bock, 1997). Similar properties might account for the observed rescuing abilities of the secreted Drosophila Fas2 protein.
Alternatively, heterophilic interaction partners could be postulated. For the Fas2 ortholog NCAM such heterophilic interaction partners have been described already, including the Ig-domain cell adhesion molecules L1 and TAG-1 and signaling receptors such as the FGF receptor and the GDNF and GDNF family receptor α (Kiselyov, 2008; Maness and Schachner, 2007; Paratcha et al., 2003). Heterophilic interaction partners are also known for the Drosophila Fas2 protein. All Fas2 isoforms share a common extracellular domain comprising five Ig domains and two fibronectin type III domains, with only isoform Fas2PF being distinct by a small exon coding for just 15 amino acids that are added to the N-terminal first Ig domain. Genome-wide interaction studies have demonstrated that the extracellular domain of Fas2 shows homophilic interactions and also binds additional Ig-domain proteins (CG33543 and CG15630 or factor of interpulse interval, fipi) (Özkan et al., 2013). We do not favor a role of CG33543 and fipi as crucial Fas2 receptors, because both proteins, which carry three Ig domains and one fibronectin type three domain, lack a discernible membrane anchor and are thus likely to be secreted proteins. Although CG15630 expression shows some overlap with Fas2 expression (Fig. S5), two mutator piggyback transposon insertions in the first coding intron, causing translational stops in all orientations, are homozygous viable (Schuldiner et al., 2008). Likewise, a CRISPR/Cas9-induced CG33543 null mutant that we generated is homozygous viable. Therefore, additional interaction partners will need to be identified.
MATERIALS AND METHODS
Genetics
Flies were raised at 25°C according to standard procedures. Rescue and gain-of-function studies were carried out using the Gal4/UAS system (Brand and Perrimon, 1993). The following fly strains were used: Fas2EB112 (Grenningloh et al., 1991); Fas2CB03613; Fas2GFP397; Fas2GFP778 (Halberg et al., 2016; Silies and Klämbt, 2010); PBac{602.P.SVS.-1}Fas2CPTI000483 [Kyoto Stock Center (DGRC) #115501]; PBac{681.P.FSVS.-1}Fas2CPTI001279 (Kyoto DGRC #115144); Mi{MIC}Fas2MI12989 [Bloomington Drosophila Stock Center (BDSC) #58577]; Mi{MIC}fipiMI06000 (BDSC #43778); Dp(1;3)DC075 (BDSC #30247); Dp(1;3)DC115 (BDSC #31445); Dp(1;3)DC465 (BDSC #32299); nsyb-Gal4, elav-GAL4 (Luo et al., 1994); repo-GAL4 (BDSC #7415); repo4.3-GAL4 (Lee and Jones, 2005); P{UAS-Stinger}2 (BDSC #65402); y1 P(nos-cas9, w+) M(3xP3-RFP.attP)ZH-2A w* (Port et al., 2014). Transgenes were generated using øC31-based transformation in the landing site 86Fb using standard protocols (Bischof et al., 2007). The Dp(1;3)Fas2 was inserted in landing site VK33 by BestGene.
The deletion alleles Df(1)Fas2 and Df(1)Fas2ΔTM were generated via FRT/Flp-mediated recombination using {XP}d07035 and PBac {WH}f06654 or PBac {PB}c04606, respectively (Exelixis collection at Harvard Medical School; Parks et al., 2004) (see also Fig. 3). In Df(1)Fas2, approximately 75 kb of the Fas2 gene locus, including the 5′ UTR, coding exons, 3′ UTR and the neighboring gene GlcAT-I were deleted. In Df(1)Fas2ΔTM, approximately 7 kb of the Fas2 3′ end, including exons coding for Fas2TM, the Fas2 3′ UTR and GlcAT-I were flipped out. Both deletions were verified by loss of white expression and/or PCR using the following primers: XP5′ hybrid AATGATTCGCAGTGGAAGGCT, WH5′ hybrid GACGCATGATTATCTTTTACGTGAC (Parks et al., 2004).
Fas2MI12989::mCherry and FipiMI06000::mCherry were generated according to standard protocols using the vector pBS-KS-attB1-2-PT-SA-SD-1-mCherry (DGRC #1307) (Venken et al., 2011).
The following crosses were performed for neuronal rescue experiments: Fas2EB112/FM7; UAS-Fas2PD-YFP/UAS-Fas2PD-YFP×w1118/FM7; nsyb-Gal4. For glial rescue experiments we used: Fas2EB112/FM7; UAS-Fas2extra-YFP/CyO×w1118/FM7; repo-Gal4/repo-Gal4; repo-Gal4/TM6. More than 500 F1 flies were analyzed for each cross. Phenotypic rescue was noted when males appeared that carried a mutant Fas2 X chromosome, which normally results in lethality. The same principle applied for mutant Fas2 rescue experiments using molecularly defined chromosomal duplications. The following crosses were performed to test the genetic requirements of Fas2: Fas2EB112/FM7 virgins were crossed to Dp(1;3)DC075; Dp(1;3)DC115; Dp(1;3)DC465; Dp(1;3)Fas2 and w1118 males. Percentages of eclosed Fas2 males and F1 flies analyzed are stated in brackets for each duplication used: Dp(1;3)DC075 (10% Fas2 males eclosed, F1 200 flies), Dp(1;3)DC115 (40% Fas2 males eclosed, F1 450 flies), Dp(1;3)DC465 (55% Fas2 males eclosed, F1 420 flies), Dp(1;3)Fas2 (52% Fas2 males eclosed, F1 650 flies), and w1118 (3% Fas2 males eclosed, F1 160 flies).
Generation of CRISPR Fas2 isoform-specific mutants
CRISPR/Cas9-based manipulations were carried out as described (Bassett et al., 2013; Port et al., 2014, 2015). sgRNAs were designed using CRISPR Optimal Target Finder with a guide length of 16-20 nt, the stringency set to ‘high’, and ‘NGG only’ for PAM site identification (Gratz et al., 2014). Chosen sgRNA had no predicted off-targets. Target sequences for isoform-specific Fas2 mutants were: Fas2PB ggtgcagcgccactcgtcgaGGG and Fas2PC ggttcagctaataacaatctCGG, which are upstream of the potential GPI anchor or predicted transmembrane domain, respectively (see Fig. S6 for details). Targeting sgRNAs were produced by in vitro transcription using the Megascript T7 Kit (Ambion) and injected into nos-Cas9 (X) flies. In Fas2ΔPB, a 28 bp deletion caused a frameshift and introduced an early translational stop after three amino acids, truncating the original Fas2PB protein sequence after 741 amino acids. Both putative GPI anchor sites (amino acid 742 and 744) were deleted. In Fas2ΔPC, a 5 bp deletion introduced a premature translational stop after eight amino acids, truncating the original Fas2PC protein sequence after 742 amino acids.
Generation of Fas2ΔPB, ΔTM double mutant
The target sequences Fas2TM U6.3 tgctgcatcaccgtccacatGGG and Fas2PB U6.1 aatccccatccctcgacgagTGG were cloned into pCFD4 plasmid and inserted in the insertion site attP86FbRFP− on the third chromosome. The target sequence for mutating Fas2TM was expressed under the U6:3 promoter, whereas the target sequence and the gRNA core sequence for Fas2PB was expressed under the U6:1 promoter. nos-cas9 flies were crossed to the resulting transgene to induce CRISPR/Cas9-mediated mutations. Sequencing of PCR products of the targeted sites revealed that 1 bp was deleted at the target site Fas2TM U6.3, and 14 bp were deleted and an additional 9 bp were inserted at the target site Fas2PB U6.1. Both mutations induced a frameshift and resulted in a termination of the Fas2TM protein translation after an additional 30 amino acids for the Fas2PC isoform, and nine amino acids for the Fas2PB isoform, respectively. Induced mutations in the allele Fas2ΔPB, ΔTM, Fas2 truncated the original Fas2TM protein sequence after 777 amino acids and the original Fas2PB protein sequence after 742 amino acids. The Fas2PB isoform lost the GPI anchoring and the Fas2TM lost part of the transmembrane domain and the entire intracellular domain. The transmembrane domain is predicted to be encoded by amino acids 759-781. The major putative GPI anchor site is localized at the serine at amino acid 744. Homozygous flies for Fas2ΔPB, ΔTM alleles were viable and fertile, and expression of the isoform Fas2TM-specific mAb 1D4 epitope was absent from all homozygous tissues tested, confirming the loss of all isoforms carrying the cytoplasmic domain.
Generation of Fas2ΔTM and Fas2ΔPB in a Fas2GFP397 background
Transgenic animals expressing a gRNA construct targeting Fas2TM (target site: aggaattgacgtcatccaagTGG) or Fas2PB (target site: aatccccatccctcgacgagTGG) were cloned into pCFD3-U6:3 and integrated into the landing site attP86FbRFP− (Bischof et al., 2007) in a Fas2GFP397 background. To introduce CRISPR-induced mutations, Fas2GFP397;;pCFD3-U6:3Fas2TM and Fas2GFP397;; pCFD3-U6:3Fas2PB transgenic males were crossed to nos-Cas9 (X) virgin flies. For Fas2GFP397, Fas2ΔTM, a 6 bp deletion and a 1 bp insertion induced a premature translational stop of the Fas2TM protein after two amino acids. The Fas2ΔTM allele results in a truncated protein after 740 amino acids, deleting the transmembrane domain and the cytoplasmic domains of Fas2TM. For Fas2GFP397,Fas2ΔPB, a 2 bp deletion induced a frameshift and a delayed translational stop of the Fas2PB protein after thirteen amino acids. The mutation results in the loss of Fas2PB GPI anchoring to the membrane and a potentially secreted protein version that is truncated after 742 amino acids. Effects on Fas2TM expression were confirmed by 1D4 epitope expression in Fas2ΔTM or Fas2ΔPB homozygous flies.
Molecular biology
Cloning of Fas2PC expression tools
pBS-KS-attB1-2-SV40-FRT backbone vector: pUAST-attB-rfA was used as template to amplify SV40 with corresponding primer PstI_SV40_for (agtctgcagggatctttgtgaaggaacct) and SV40_EcoRI_rev (CGCAGAATTCggatccagacatgataagat) introducing PstI and EcoRI restriction sites (underlined). PstI-SV40-EcoRI PCR product was digested with PstI/EcoRI and purified using column purification (Qiagen). PstI/EcoRI-digested SV40 PCR product was ligated into PstI/EcoRI-cleaved pBS-KS-attB1-2-FRT vector.
pBS-KS-attB1-2_Fas2PC::V5_Fas2PB::HA_SV40_FRT_white: V5 and HA tags were introduced using RF (restriction free) cloning. BsaI restriction sites were added to the different fragments (fragment 1: Fas2PB::HA; fragment 2: Fas2PC::V5) during amplification from w1118 gDNA. BsaI-digested fragments 1+2 were ligated in one reaction into the XbaI/SpeI-digested backbone vector pBS-KS-attB1-2-SV40-FRT. Primers used for RF cloning carried additional nucleotides encoding the amino acid glycine (indicated by lower-case letters): RF_cloning_Fas2PC::V5_for (CCACCACAATAAGCATAACATTACTTAGTGTCCTAGCCTCAATGTTAGCCggaGGTAAGCCTATCCCTAACCCTCTCCTCGG), RF_cloning_Fas2PB::HA_for (TATCGAAAATCGA CAACCAATATCGAAAAAAACCAACAGATAATCCCCATggaTACCCATACGATGTTCCAGATTA), RF_cloning_Fas2PB::HA_rev (GTGAAAATTACAAGCAGTTGGGCCAGGGGTGCAGCGCCACTCGTCG AGGGtccAGCGTAATCTGGAACATCGTATG), RF_cloning_Fas2PC::V5_rev (tgtcacaccacagaagtaaggttccttcacaaagatccctgcagactagtttatccCGTAGAATCGAGACCGAGGAGAGGGTTAGGG). Lower-case letters indicate end of vector and added SpeI restriction site. Primers used to amplify fragments 1 and 2, and to clone into pENTR, introducing restriction sites (as indicated by lower-case letters) and leaving XbaI (for Fragment 2), and SpeI (for Fragment 1) overhangs for subsequent cloning were: BsaI_XbaI_MI12989_Fas2PC_for_1 (CACCggtctctctagACCTAGCTCAGGAATTTGTT), MI12989_Fas2PC_BsaI_rev_1 (ggtctcaTTAGGCTAACATTGAGGCTA), BsaI_MI_Fas2RC_RNA_for_2 (CACCggtctcaCTAAGCATTACAATTCGTTC), MI_Fas2PB_SpeI_BsaI_rev_2 (ggtctcactagTTAAGCAGTGTGCGTCGTCG). At the end, PCR-amplified white marker was introduced into all of the plasmids via XhoI/HindIII digest. All plasmids were injected into Mi{MIC}Fas2MI12989 (BDSC# 58577) flies to generate transgenes.
Generation of Fas2 constructs
UAS constructs of Fas2PB and Fas2PC CDS were generated using standard procedures, w1118 embryonic cDNA as template, and a pUAST-attB-rfA plasmid as destination vector. Primers used for cloning: TOPO Fas2 Exon1 for caccATGGGTGAATTGCCGCCAAA and Fas2 PB cDNA rev TTAAGCAGTGTGCGTCGTCG or Fas2 PC cDNA rev TTAGGCTAACATTGAGGCTA respectively. Site-directed mutagenesis (NEB) was used to introduce a single HA tag [GGYPYDVPDYAGG, with two adjacent glycins as spacers (underlined)] at position +100 bp of the Fas2PB and Fas2PC CDS, directly after the endogenous signal peptide sequence (UAS-Fas2HA-PB and UAS-Fas2HA-PC). Primers used for site-directed mutagenesis were: Fas2HA for (CTCTGCAGCTGCTCTTTAATAGAACTGACCCGTGCGCAGTCCCCCATCCTGggaggaTACCCATACGATGTTCCAGATTAC) and Fas2HA rev (CATACGATGTTCCAGATTACGCTggaggaGAGATTTATCCCAAACAAGAAGTCCAGCGCAAGCCAGTGGGCAAGCCCCTG). Lower-case letters indicate insertion of a linker sequence encoding two glycine residues.
RNA extraction and RT-PCR
RNA was isolated from w1118 embryos (all stages) using TRIsure (Bioline) and RT-PCR was performed using SuperScript II Reverse Transcriptase (Invitrogen) according to manufacturer's guidelines. Oligonucleotides used for detection of Fas2 splice variants were: Fas2 A for (CTGAGCGACAGGTCTTCTCC), Fas2 A rev (GCCGAATTCTTCCCGATTAT), Fas2 A PEST rev (CGGTGGCTCCTTTACCAG), Fas2 Exon6 for (CAGATTCGCCGTACATTGTG), Fas2 B rev (CAAAATCAGCAGCATTGTCG), Fas2 C rev (TGTGGCTGTTGTTGTTGTTG), Fas2 F for (CGATCCACTGTATGATTCCAAG), Fas2 F rev (GTGGCATCTCGAATCCAACT), Fas2-RG for (TTTACGGTTGGCGTTTTCCG), Fas2-RH rev (CCGCGCATAAACCCAATTGT).
Immunohistochemistry and live imaging
Fixation and preparation of tissues for immunohistochemistry was performed as described previously (Yuva-Aydemir et al., 2011). Antibodies used were: mouse anti-Repo (8D12, 1:5), mouse anti-Fas2 (1D4, 1:5) (all Developmental Studies Hybridoma Bank); rabbit anti-GFP (A-11122, 1:1000; Invitrogen); anti-dsRed (632496, 1:1000; Takara Bio) and goat anti-HRP-DyLight 647 conjugated (123-605-021, 1:1000; Dianova). Secondary antibodies used in this study were: goat anti-mouse IgG and goat anti-rabbit IgG with the fluorophore conjugates Alexa Fluor 488, 568 and 647 (488: mouse A1101, rabbit A11008; 568: mouse A11004, rabbit A11011; 647: mouse A21235, rabbit A21244; all 1:1000; Molecular Probes). Specimens were analyzed using a Zeiss LSM710 or LSM880 confocal microscope. Live-imaging analysis was performed using an UltraView RS (Perkin Elmer) or a Zeiss LSM 5 Duo microscope. Original confocal data, images, orthogonal sections, and movie sequences were processed using Zeiss ZEN 2012 software (Zeiss), Adobe Photoshop CS6, and Fiji (Schindelin et al., 2012). The chorion of staged embryos was removed by 50% NaCl treatment for 3 min. Embryos were aligned on heptane-coated coverslips in glass-bottom dishes (MatTek Corporation) and imaged covered in 10S Voltalef oil to prevent drying of embryos. Glial cell migration was recorded with a speed of 1 stack/min over 120-180 min. At least nine and up to 30 embryos were imaged for each genotype.
Cell movement analysis
Manual tracking of glial cell migration was performed using Fiji (control: n=4 embryos; Fas2EB112: n=7 embryos; 30-40 individual glial cell nuclei were tracked per movie). In order to quantify aberrant cell movements, we implemented a custom Matlab script. First, we extracted the movement direction using consecutive positional tracking estimates of the nuclei. As shown in Fig. S4I, the movement is characterized by multiple cell clusters migrating on parallel lines from a common starting line (line A in Fig. S4I) to a shared destination line (line B in Fig. S4I). Using the direction of a vector orthogonal to these lines as the standard movement direction (i.e. the mode direction), we then quantified nuclei motion by counting the number of frames in which the movement direction deviates more than ±90° of this mode direction (i.e. moving in the opposite direction of the mode; compare with Fig. S4J). By incrementing a counter every time a misdirection movement event (MME) is detected we can plot the length of the MMEs against the overall count of frames with misdirected movement (Fig. S4K).
Statistical analysis of Fas2 intensity ratio
Statistical analysis of Fas2 intensity ratio was performed as previously published (Silies and Klämbt, 2010). In short, to quantify Fas2 expression, two regions of interest close to the CNS-PNS transition zone and in a distal region of the nerve behind the growth cone were defined and the mean intensity was determined using ImageJ software (NIH). Fas2 expression in nerves of two to six hemisegments per embryo was measured. The mean of Fas2 intensity ratio per embryo was used for subsequent statistical analysis using the Mann–Whitney Rank Sum test.
Biochemical methods
UAS-Fas2PB and UAS-Fas2PC constructs were transfected in S2R+ cells (Stork et al., 2009). After 2 days, cells were incubated for 1 h at 25°C with serum-free medium, or serum-free medium containing phosphatidylinositol-specific phospholipase C (PiPLC, Sigma-Aldrich, P5542) (1 U/ml) as described (Petri et al., 2019). Isolated proteins from cell lysates and supernatant were analyzed by western blot.
Acknowledgements
We are thankful to all members of the Klämbt lab for help throughout the project and comments on the manuscript; J. Börgers for help during the generation of the mCherry conversion of the Fas2 MiMIC insertion; and F. Port for sharing of CRISPR reagents and much advice.
Footnotes
Author contributions
Conceptualization: H.N., M.S., C.K.; Methodology: H.N., P.D., K.K., E.N.; Software: B.R.; Investigation: H.N., P.D., K.K., E.N., G.S., B.R., M.S.; Writing - original draft: C.K.; Writing - review & editing: C.K., H.N., M.S.; Visualization: H.N., C.K., B.R.; Funding acquisition: C.K.
Funding
Work in the Klämbt lab is supported by a grant from the Deutsche Forschungsgemeinschaft (DFG) (SFB 1348 B5).
References
Competing interests
The authors declare no competing or financial interests.