The unc-5 locus is predicted to encode at least seven isoforms, A-C, D1, D2, E and F (Wormbase WS245) (Figure 1B). Six of the seven isoforms (A, C-F) differ by exon usage at the 5′ end of the gene, resulting in distinct extracellular N-termini. unc-5B differs at the 3′ end from the other 5 isoforms, resulting in a truncated 3′ end (Figure 1B). The short isoform unc-5B is generated by skipping the 5’ splice donor site of intron 7 (second to the last intron) (Figure 1C). The open reading frame extends 44 bases into intron 7 before encountering a stop codon (Figure 1C), resulting in an additional 15 amino acid residues with no similarity to other known sequences. In RNA-seq of mixed stage animals, 242 reads contained the exon 7 to exon 8 splice, and 9 reads overlapped exon7/intron 7 or intron 7/exon 8 boundaries (Figure 1D). Neighboring introns 6 and 9 showed no reads overlapping the exon/intron boundaries, suggesting that the intron 7 reads are representative of unc-5B. These results indicate that unc-5B constituted 3.6% (9/251) of unc-5 transcripts in mixed-stage animals.

The UNC-5A long isoform encodes a molecule with two extracellular immunoglobulin (Ig) domains, two extracellular thrombospondin type I repeats (TSPI), a transmembrane domain, and cytoplasmic zona occludens I/UNC-5 domain (ZU5), UNC- 5/PIDD/Ankyrin (UPA), and DEATH domains (Figure 2A). A tyrosine residue (Y482) that is phosphorylated and is important for functional UNC-5 (Killeen et al., 2002) is found in the intracellular juxtamembrane region (Figure 2A). The short isoform UNC-5B lacks the DEATH and UPA domains and most of the ZU5 domain, but retains Y482 (Figure 2A).

Vertebrates, including humans, have four unc5 genes, A-D. The human Unc5A locus encodes a truncated isoform similar to C. elegans UNC-5B, called Unc5A X4 (nucleotide XM_011534687.2, protein XP_011532989.1) (Figure 2B). Unc5A X4 also includes a portion of the ZU5 domain before ending with a stop codon (Figure 2B). unc5A X4 isoforms are also predicted in other vertebrate species (e.g., mouse and Xenopus).

The unc-5 isoforms A, C, D, E and F are collectively referred to as unc-5 long isoforms, and unc-5B the short isoform. The unc-5 alleles e791, e53, e553, and e152 introduce premature stop codons that affect all unc-5 isoforms (Figures 1A, 2). The ev480 and op468 alleles affect only the unc-5 long isoforms and do not affect the unc- 5B short isoform (Figures 1A, 2). ev480 introduces a premature stop codon, and op468 is a 337-bp out-of-frame deletion. The e791 and e53 mutations predicted to affect all isoforms caused a strong uncoordinated locomotion (Unc) phenotype (Table 1). The long-isoform-specific mutations ev480 and op468 caused a weaker Unc phenotype (Table 1). The e152 allele, which also is predicted to affect all isoforms, also caused a weaker Unc phenotype (Table 1). Based on axon guidance defects, e152 is also a hypomorphic mutation (Table 1), similar to previous findings (Killeen et al., 2002).

VD/DD axon guidance defects were scored in these mutants. The 16 VD/DD commissures, composed of 19 axons, extend directly ventrally to dorsally (Figure 1B; 3A). In Figure 3, both VD and DD axons are shown. e791 and e53 mutants displayed very severe VD/DD guidance defects, with between only 5%–7% of axons extending dorsally past the lateral midline of the animal (Table 1; Figure 3B). In contrast, ev480, op468, and e152 displayed significantly less severe defects, with between 70%–85% extending past the lateral midline (Table 1; Figures 3C, D). Despite extending past the lateral midline, most axons in these mutants were still misguided as evidenced by lateral wandering (Figures 3C, D). These data indicate that ev480, e152, and op468, which only affect the long isoforms, are hypomorphic, incomplete loss-of-function mutations.

The hypomorphic e152 mutation resides in exon 7 (Figure 1C), which is included in all six isoforms, yet e152 has a hypomorphic phenotype (Table 1; Figure 3). Exon 7 is the terminal exon in unc-5B short, and the e152 mutation resides 183 nucleotides (61 codons) from the end of the predicted unc-5B reading frame. This is downstream of the transmembrane domain and Y482. Any truncated molecule made by translation of e152 would include both (Figure 1). Most of the truncated ZU5 domain would also be included. Possibly, unc-5B short function is retained in e152 by bypassing nonsense-mediated decay due to its location near the stop codon in the terminal exon. However, axon guidance defects in e152 are significantly stronger than ev480 and op468 (Table 1), suggesting that e152 might be having some effect on unc-5 long isoforms, just not complete elimination of activity. In sum, these data showing that long-isoform-specific mutations are hypomorphic suggest that the UNC-5B short isoform has a role in axon guidance.

During dorsal migration, VD growth cones display a robust lamellipodial growth cone body with filopodial protrusions biased to the dorsal region of the growth cone (Knobel et al., 1999; Norris and Lundquist, 2011) (Figures 4A–D). The strong loss-of-function unc-5(e53) displayed growth cones that were excessively protrusive compared to wild-type, with larger growth cone area and longer filopodial protrusions (Norris and Lundquist, 2011). Dorsal bias of filopodial protrusion was also lost in unc-5(e53) (Norris and Lundquist, 2011). Strong loss-of-function unc-5(e791) showed a similar loss of filopodial dorsal polarity (Figure 4C) and excess protrusion, as evidenced by increased growth cone area (Figure 4A) and increased filopodial length (Figure 4B). The unc-5(e152) hypomorphic allele also displayed increased growth cone area and filopodial length, and loss of dorsal bias of filopodial protrusion (Norris and Lundquist, 2011) (Figure 4), as did the unc-5(ev480) hypomorphic mutant (Figure 4).

F-actin was visualized using a transgene that expresses a GFP-tagged version of the F-actin binding domain of VAB-10 (see Materials and Methods) (Norris and Lundquist, 2011; Norris et al., 2014; Gujar et al., 2018). Previous studies showed that F-actin in VD growth cones shows a dorsal bias of F-actin accumulation that is dependent upon UNC-5 (Norris and Lundquist, 2011; Norris et al., 2014; Gujar et al., 2018). Microtubule + ends were visualized using a transgene expressing a GFP-tagged version of EBP-2, a microtubule + end binding protein (Gujar et al., 2018). UNC-5 was required to restrict the presence of microtubule + ends in growth cones, which have a pro-protrusive effect (Gujar et al., 2018). These studies were conducted with strong loss-of-function alleles of unc-5.

Hypomorphic unc-5(e152) and unc-5(ev480) also displayed loss of dorsal F-actin accumulation (Figures 5A, D) and increased numbers of microtubule + ends in growth cones (Figures 5E, G). In sum, these results indicate that the UNC-5 long isoforms are required for growth cone dorsal polarity of filopodial protrusion and F-actin accumulation, to inhibit growth cone protrusion, and to restrict MT + end entry into the VD growth cone.

A mutation predicted to affect only the unc-5B short isoform was constructed by CRISPR/Cas9 genome editing. unc-5B(lq190) was produced by precise removal of intron 7, which is predicted to preclude the formation of the unc-5B short isoform by fusing together exons seven and eight (Figure 1B), leaving the long isoforms unaffected.

unc-5B(lq190) mutants displayed wild-type locomotion and weak but significant total axon defects (3%) including axon wandering (Table 1; Figure 3K). unc- 5B(lq190) did not significantly affect VD growth cone area (Figure 6A). However, filopodial length was reduced compared to wild-type (Figure 6B), and dorsal polarity of filopodial protrusion was abolished (Figures 6C, D). Consistent with the loss of dorsal filopodial polarity, F-actin dorsal accumulation was also lost in unc-5B(lq190) (Figures 5A, D). Microtubule + end accumulation was unaffected (Figures 5E, H).

To ensure that the phenotype observed in unc-5B(lq190) was due to the unc-5 genome edit and not to a background mutation, we analyzed trans-heterozygotes of unc-5B(lq190) and the strong loss-of-function allele unc-5(tmIs1241) found on the balancer chromosome tmC25 (Dejima et al., 2018). tmIs1241 is an insertion of a myo- 2::gfp transgene in the unc-5 exon seven shared by long and short isoforms (Dejima et al., 2018), and has a strong Unc phenotype and severe VD/DD axon guidance defects similar to unc-5(e791) (data not shown). Thus, unc-5(tmIs1241) a strong loss-of-function unc-5 mutant. VD growth cones of lq190/tmC25 trans-heterozygotes resembled unc- 5B(lq190) alone, with significantly reduced filopodial length compared to wild-type and loss of dorsal polarity of filopodial protrusion (Figure 6). This indicates that unc- 5(tm1241) failed to complement unc-5B(lq190) as expected.

An lq190 trans-heterozygote with the hypomorphic ev480 allele displayed filopodial length similar to wild-type and dorsally-polarized filopodial protrusion (Figure 6). Thus, unc-5(ev480) complemented unc-5B(lq190), consistent with the idea that these isoforms encode distinct functions. In sum, these results indicate that unc-5 long and unc-5B short encode genetically distinct and separable functions. unc-5B short isoform is required for dorsal polarity of filopodial protrusion, as well as robust filopodial protrusion, suggesting a pro-protrusive role.

The phenotypic analysis above suggests that the UNC-5 long isoforms and the UNC-5B short isoform are each involved in VD/DD axon guidance. An unc-5::gfp minigene is composed of the unc-5 promoter driving the unc-5A cDNA fused to gfp (Killeen et al., 2002). This unc-5A minigene lacks introns and therefore is predicted to be unable to produce the unc-5B short isoform. Transgenic expression of unc-5A significantly rescued the Unc phenotype, VD/DD lateral midline crossing axon defects, and total VD/DD axon guidance defects of unc-5 mutants, including e791 and hypomorphic e152 and ev480 mutants (Table 1; Figures 3G, I). unc-5A expression in a wild-type background resulted in a slightly uncoordinated phenotype with weak but significant VD/DD axon guidance defects (Table 1; Figure 3E). This is possibly due to transgenic overexpression.

An unc-5B short isoform minigene was constructed by deleting exons eight and nine of the unc-5A minigene and replacing these sequences with the first 44 bases of intron 7 encoding the C-terminal 15 residues of UNC-5B short, followed by gfp (Figure 1C). This unc-5B minigene has the capacity to produce a molecule with the UNC-5B C-terminus, and not the longer isoforms. Transgenic unc-5B in a wild-type background also led to weak but significant VD/DD axon guidance defects and a slightly Unc phenotype (Table 1; Figure 3F). unc-5B rescued the Unc, midline crossing, and total axon guidance defects of e791, e152, and ev480. However, rescue of total axon guidance defects was significantly weaker than unc-5A (Table 1; Figures 3G–J). These data indicate that UNC-5A long and UNC-5B short can both facilitate VD/DD axon guidance.

A constitutively-active UNC-5A transgene was produced by adding an N-terminal myristoylation sequence to the cytoplasmic domain of UNC-5A (Norris et al., 2014), identical to what was done to constitutively activate the UNC-6/Netrin receptor UNC-40 (Gitai et al., 2003). In a manner similar to receptor tyrosine kinases, deletion of the extracellular domain results in constitutive activation (Ullrich and Schlessinger, 1990). The extracellular domains of dimerized receptors are thought to inhibit interaction of the cytoplasmic domains, which is relieved by ligand binding. Thus, MYR::UNC-5 likely has constitutive cytoplasmic domain interaction and thus constitutive activation. The phenotype of MYR::UNC-5 expression in the VD growth cone is the opposite of unc-5 loss-of-function, consistent with constitutive activation. To wit, expression of MYR::UNC-5A in the VD/DD neurons strongly inhibited VD growth cone area and filopodial protrusion length, consistent with a role of UNC-5A in inhibiting protrusion (Norris et al., 2014) (Figures 7A, B). MYR::UNC-5A expression caused no significant defects in polarity of protrusion (Figure 7C).

A MYR::UNC-5B short isoform transgene was constructed with the 3’ end of unc- 5B (Figure 2A). Expression of MYR::UNC-5B in VD/DD neurons caused no significant change in growth cone area, filopodial length, or polarity (Figure 7). This is consistent with unc-5B(lq190) short isoform having a distinct role from the long isoforms, which inhibit protrusion.

The UNC-6/Netrin receptor UNC-40/DCC has a dual role in controlling growth cone protrusion. It acts as a heterodimer with UNC-5 to inhibit protrusion, and as a homodimer to stimulate protrusion (Norris and Lundquist, 2011; Norris et al., 2014; Gujar et al., 2018). unc-40(n324) null mutant displayed a slight but not significant reduction in growth cone area, and significantly shorter filopodial protrusions (Figures 8A, B, D). unc-40(n324) did not significantly affect dorsal polarity of filopodial protrusions (Figure 8C). This intermediate phenotype likely reflects the dual role of UNC-40 in stimulation and inhibition of VD growth cone protrusion.

Previous studies showed that UNC-40 function was required for the excess growth cone lamellipodial and filopodial protrusion seen in unc-5 strong loss-of-function mutants that affect both long and short isoforms. Thus, in an unc-5 mutant, the pro-protrusive activity of UNC-40 was overactive, resulting in excess protrusion (i.e., unc-40 suppressed unc-5) (Norris and Lundquist, 2011). unc-5(lq190); unc-40(n324) double mutant growth cones resembled the additive effect of each mutant alone: slight but not significant reduction in growth cone area; significantly reduced filopodial length, and loss of dorsal polarity of filopodial protrusion (Figure 8). This suggests that UNC-5B short isoform and UNC-40 might act in the same pathway for filopodial length. However, unc-5B(lq190) synergistically increased unc-40(n324) VD/DD axon guidance defects (Table 1; Figure 3M), suggesting that they might act in parallel in other aspects of growth cone morphology not assayed here.

unc-6(ev400) strong loss-of-function resulted in unpolarized growth cones with no significant effect on growth cone and filopodial protrusion (Figure 8). unc-6(ev400); unc-5(lq190) growth cones showed significantly reduced growth cone area compared to either single mutant alone. Filopodial length resembled the short filopodia of unc- 5(lq190) alone. This suggests that unc-6 and unc-5B short might act in parallel to promote growth cone protrusion and is consistent with a pro-protrusive role of unc-5B short isoform.

unc-5 hypomorphic mutants that affect only the long isoforms also displayed excess growth cone protrusion (Figure 4). unc-40(n324) did not significantly reduce increased growth cone area or filopodial length in unc-5(ev480) hypomorphic mutant (Figures 8A, B, F). This suggests that the excess protrusion seen in the unc-5(ev480) hypomorph is not solely due to overactivity of UNC-40. Possibly, the UNC-5B short isoform has pro-protrusive activity that is increased in the absence of the long isoforms. This is consistent with unc-5B(lq190) short isoform mutants having reduced filopodial length.

Previous results showed that unc-40 mutation suppressed the VD/DD axon guidance defects of unc-5(e53) strong loss-of-function and of unc-5(e152) hypomorphic mutants (Norris and Lundquist, 2011). In contrast, the unc-5(ev480) hypomorph was not suppressed by unc-40(n324) and in fact showed synergistic enhancement for failure of lateral midline crossing and total axon guidance defects (Table 1; Figure 3N). This is consistent with an UNC-40-independent mechanism in unc-5(ev480) hypomorphic mutants driving excess protrusion, possibly involving UNC-5B.

Transgenic expression of UNC-5A and UNC-5B each rescued the uncoordinated locomotion and VD/DD axon guidance defects on unc-5(e791) mutants (Table 1). UNC-5A expression in a wild-type background did not affect growth cone area or filopodial length (Figures 9A, B), but did significantly reduce dorsal polarity of filopodial protrusion, but not to the extent of unc-5 mutants (Figure 9C). This might explain in part the axon guidance defects caused by unc-5A expression in a wild-type background (). UNC-5A expression significantly rescued excess growth cone area, filopodial length, and dorsal polarity of filopodial protrusion in unc-5(e791) mutants (Figures 9A–C).

UNC-5B expression in a wild-type background caused significantly increased growth cone area, but did not affect filopodial length (Figures 9A, B). Furthermore, UNC-5B expression abolished dorsal polarity of filopodial protrusion (Figure 9C). This might in part explain the VD/DD axon guidance defects caused by unc-5B expression (Table 1).

UNC-5B expression failed to significantly rescue growth cone area or dorsal polarity of filopodial protrusion in unc-5(e791) mutants (Figures 9A, B), but did significantly reduce filopodial length (Figure 9C). It is possible UNC-5B acts as a dominant-negative for UNC-5A activity. However, UNC-5B expression rescued uncoordinated locomotion and filopodial length of unc-5 mutants, suggesting that it might have roles in the VD growth cone distinct from UNC-5A.

Tyrosine 482 was previously shown to be required for the ability of an unc-5 transgene to rescue unc-5 mutants (Killeen et al., 2002). Tyrosine 482 was mutated to non-phosphorylable phenylalanine in unc-5A and unc-5B transgenes. Rescue of unc-5(e791) VD/DD axon guidance defects by unc- 5A(Y482F) was significantly reduced compared to unc-5A(+) (Table 1).

In a wild-type background, unc-5A(Y482F) caused a significant increase in growth cone area and filopodial length similar to unc-5(e791) mutants (Figures 9A, B), and abolished dorsal polarity of filopodial protrusion (Figure 9C). Furthermore, it failed to significantly rescue growth cone area, filopodial length, and polarity of protrusion in unc-5(e791) mutants (Figure 9). These data indicate that Y482F is required for UNC-5A function, and that UNC-5A (Y482) might have a dominant negative effect on wild-type UNC-5A.

unc-5B(Y482F) failed to rescue VD/DD axon guidance defects in unc-5 mutants, indicating that Y482 is important in UNC-5B function (Table 1). It also failed to rescue filopodial length defects of unc-5 mutants (Figure 9B). UNC-5B(Y482F) expression alone still caused an increase in growth cone area and abolished dorsal polarity of filopodial protrusion (Figures 9A, C). This suggests that Y482 is required for some aspects of UNC-5B function (axon guidance and filopodial length) but not others (increased growth cone area and disruption of dorsal polarity of protrusion.

unc-5 mutations that affect all unc-5 isoforms (e.g., e791) caused severely uncoordinated locomotion and severe VD/DD axon guidance defects (unc-5 strong loss-of-function). Very few axons were observed emanating from the ventral nerve cord, and those that did failed to extend dorsally past the lateral midline (Figure 3B; Table 1). unc-5 mutations that affect only the long isoforms were less severely uncoordinated than unc-5 strong loss-of-function mutants, and displayed less-severe VD/DD axon guidance defects (unc-5 hypomorphs) (Table 1; Figure 1). However, VD growth cone morphology of unc-5 hypomorphs was not significantly different from that of unc-5 strong loss-of-function. They were unpolarized with larger growth cones and longer filopodial protrusions, and displayed unpolarized F-actin and excess microtubule + ends. It is likely that subtle differences in VD growth cone morphology between strong and hypomorphic unc-5 mutants were not detected in these assays. In any case, the weaker uncoordination and weaker axon guidance defects of unc-5 hypomorphs imply that the unc-5B short isoform has a role in VD/DD axon guidance.

A mutation precited to specifically affect the unc-5B short isoform was generated by CRISPR/Cas9 genome editing. The unc-5B 3’ end results from failure to splice intron 7 from the unc-5 transcript, resulting in a stop codon after 16 additional amino acid coding codons (Figure 1). RNA-seq revealed unc-5B reads in intron 7 (Figure 1). unc-5B(lq190) was a precise removal of intron 7, resulting in the fusion of exons 7 and 8, leaving unc-5 long isoform coding potential unchanged. unc-5b(lq190) mutants were only slightly uncoordinated with weak VD/DD axon guidance defects (Table 1; Figure 3). However, unc-5B VD growth cones were unpolarized and displayed significantly shorter filopodial protrusions compared to wild-type (Figure 6). unc-5B(lq190) failed to complement an unc-5 strong loss-of-function mutation for these phenotypes, but did complement an unc-5 hypomorphic long isoform-specific mutant (Figure 6). Thus, the functions of unc-5 long and short isoforms are genetically distinct and separable. unc-5B(lq190) resulted in unpolarized F-actin, but did not affect microtubule + end abundance (Figure 5), consistent with the growth cone phenotypes of loss of polarity but no effect on growth cone area.

These data suggest that UNC-5B short has a pro-protrusive role, in contrast to the inhibitory role of UNC-5 long. Consistent with this idea, unc-5B(lq190) double mutants with unc-6 had severely reduced VD growth cone area compared to wild-type and single mutants alone, and transgenic expression of unc-5B resulted in significantly increased growth cone area compared to wild type (Figure 9).

Transgenic expression of unc-5A long in a wild-type background caused slightly uncoordinated locomotion and weak VD/DD axon guidance defects (Table 1; Figure 3). unc-5A long expression did not significantly alter growth cone area or filopodial length, but did significantly perturb dorsal polarity of protrusion, but not to the extent of unc-5 strong loss-of-function mutants (Figure 9). unc-5A long transgenic expression did significantly rescue the uncoordinated locomotion and VD/D axon guidance defects of unc-5 strong loss-of- function mutants, as well as VD growth cone area and filopodial length increase (Table 1; Figures 3, 9). Thus, while unc-5A long transgenic expression caused defects alone, it did rescue unc-5 loss-of-function.

Transgenic unc-5B short expression in a wild-type background also caused weak uncoordinated locomotion and VD/DD axon guidance defects, as well as loss of dorsal polarity of protrusion and increased growth cone area (Table 1; Figures 3, 9). unc-5B short transgenic expression rescued uncoordinated locomotion and VD/DD axon guidance defects of unc-5 mutants, but did not rescue polarity or growth cone area defects of unc-5 strong loss-of-function. This indicates that more subtle effects of unc-5B short expression on the growth cone are not being resolved in these studies. unc-5B short expression did rescue excess filopodial length of unc-5 strong loss-of-function (Figure 9), suggesting that in some contexts UNC-5B short might also inhibit filopodial protrusion. These results are consistent with previous studies showing that unc-5 transgenes lacking cytoplasmic domains still retained function in axon guidance ( Killeen et al., 2002).

These data indicate that both unc-5A long and unc-5B short can compensate for some aspects of unc-5 loss-of function in axon guidance and that UNC-5B short has a novel role in axon guidance and growth cone morphology.

Previous studies showed that tyrosine 482 (Y482) is phosphorylated and that it is required for the function of UNC-5 in axon guidance ( Killeen et al., 2002). Studies here confirm that Y482 is required for the rescuing effects of the unc-5A long transgene. unc- 5(Y482F) does not rescue the uncoordinated locomotion or VD/DD axon defects of unc- 5 mutants as well as the wild-type transgene (Table 1). Furthermore, it fails to rescue the increased growth cone area, excessively long growth cone filopodia, and loss of dorsal polarity of protrusion of unc-5 mutants (Figure 9). Thus, Y482 is required for the function of UNC-5A long isoform.

Y482 was required for rescue of axon guidance and increased filopodial length of unc-5 mutants. However, unc-5B(Y482F) expression alone still resulted in increased growth cone area and loss of filopodial protrusion. Thus, in UNC-5B short, Y482 is required for some functions but not others. While it is unclear which tyrosine kinase might phosphorylate Y482, the SH2 domain of SRC-1 binds to phosphorylated Y482, SRC-1 is required for robust UNC-5 tyrosine phosphorylation, and SRC-1 functions with UNC-5 in gonadal distal tip cell migration and axon guidance ( Lee et al., 2005). Indeed, replacing the entire UNC-5 cytoplasmic region with SRC-1 in a fusion protein rescued unc-5 mutants ( Lee et al., 2005), suggesting that recruitment of SRC-1 is the role of the cytoplasmic domain of UNC-5. However, we show here that the UNC-5B short isoform lacking most of the cytoplasmic region of UNC-5 partially rescues unc-5 mutants, so SRC-1 might not be required.

These results indicate that the UNC-5B short isoform is required for VD growth cone dorsal polarity of filopodial protrusion and for robust growth cone lamellipodial and filopodial protrusion. This is in contrast to the UNC-5A long isoform, which inhibits growth cone protrusion. Double mutants of unc-5B(lq190); unc-40 displayed an additive phenotype, and no significant genetic interaction. This suggests that UNC-5B short might act independently of UNC-40, or in the same pathway. UNC- 5B short is lacking the cytoplasmic UPA/DB domain that mediates physical interaction with the UNC-40 P1 cytoplasmic domain. This suggests that UNC-5B short might act independently of UNC-40.

unc-5B(lq190); unc-6 double mutants displayed growth cones that were smaller than wild-type and either single mutant alone. This suggests that UNC-5B short might act in parallel to UNC-6 to promote growth cone lamellipodial protrusion. In the Polarity/Protrusion model, UNC-6 acts through UNC-40 to stimulate growth cone protrusion. However, this effect is independent of UNC-40 as unc-5B(lq190); unc-40 double mutants did not display small growth cones similar to unc-5B(lq190); unc-6 mutants. This suggests that an unidentified ligand might interact with UNC-5B short to stimulate protrusion in parallel to UNC-6, and that an unidentified receptor for UNC-6 might act in parallel to UNC-5B short (Figure 10). The identities of this potential and ligand are unknown, but could other cues and receptors that guide axons in the dorsal ventral axis and interact with UNC-6/netrin signaling, such as Slit/Robo (Tang and Wadsworth, 2014; Xu et al., 2015), Draxin (Gao et al., 2015), or dsCAM (Purohit et al., 2012).

In sum, this work describes a novel role of a short UNC-5 isoform lacking most cytoplasmic domains in axon guidance and growth cone morphology. Previous studies showed that UNC-5 homodimers and UNC-5:UNC-40 heterodimers inhibit VD growth cone protrusion, and that UNC-40 homodimers stimulate growth cone protrusion in response to UNC-6 (Figure 10) (Norris and Lundquist, 2011; Norris et al., 2014; Gujar et al., 2018; Gujar et al., 2019; Mahadik and Lundquist, 2022; Mahadik and Lundquist, 2023). These results show that the UNC-5B short isoform is pro-protrusive, acts independently of UNC-40, and in parallel to UNC-6. In human neuroblastoma SH-SY5Y cells differentiated into neurons, Unc5A was required for robust neurite outgrowth (Huang et al., 2018). While the nature of this neurite outgrowth defect is unclear, it could represent a pro-protrusive role of Unc5A in these cells, similar to the role of unc-5B described here.

Experiments were performed at 20°C using standard C. elegans techniques (Brenner, 1974). The presence of mutations was confirmed by phenotype and sequencing. The following integrated transgenes were produced, with unknown chromosomal location: lqIs283 [Punc-5::unc-5A], lqIs308 [Punc-5::unc-5B], lqIs360 [Punc-5::unc-5A(Y482F)], lqIs387 [Punc-5::unc-5B(Y482F)], lqIs326 [Punc-5::myr::unc- 5A], lqIs315 [Punc-5::myr::unc-5B].

The PUNC5GFP transgene (Killeen et al., 2002) was used as a basis to construct the transgenes described in this work. The Punc-5::unc-5B short isoform transgene was constructed by removing the coding sequence of exons eight and nine, and replacing with the first 45 bases of intron 7 (Figure 1) fused in-frame to gfp. The tyrosine 482 mutation to phenylalanine was generated by site-directed mutagenesis.

RNA-seq was conducted using total RNA isolated from three independent isolates of mixed-stage animals of the strain LE6194 (wrdSi23 I; juIs76 II). (SSM1, SSM2, and SSM3) as previously described (Tamayo et al., 2013). LE6194 is wild-type for the unc-5 gene. Stranded poly-A RNA-seq libraries were made using the NEBNext^® Ultra™ II Directional RNA Library Prep Kit for Illumina. Paired-end 150 cycle sequencing on a Nextseq550 was utilized. Reads were aligned to the C. elegans genome using HISAT2 with default settings (version 2.1.0) (Kim et al., 2019). Resulting BAM files were analyzed in the Integrated Genome Viewer (Robinson et al., 2011; Thorvaldsdottir et al., 2012) (Figure 1D). Sequences are available in the Sequence Read Archive, BioProject number PRJNA847250.

unc-5(lq190) was generated by CRISPR/Cas9 genome editing which precisely deleted the entire intron 7 of unc-5A, between exon 7 and 8 (Figure 1). This removes the truncated 3′end of UNC-5B, which resides in intron 7, and leaves the coding potential of unc-5 long isoforms unchanged. Synthetic guide RNAs were directed at the 5′ and 3′ ends of intron 7. A mix of sgRNAs, a single stranded oligonucleotide repair template with recoded sgRNA regions, and Cas9 enzyme was injected into the gonads of N2 animals, along with the dpy-10(cn64) co-CRISPR reagents (EL MOURIDI et al., 2017). Deletion of unc-5 intron 7 was confirmed by PCR and sequencing and outcrossed to N2 three times. Genome editing reagents were produced by In Vivo Biosystems (Eugene, OR, USA):

sgRNA1: 5′ CTC​TCC​CCA​GTC​ATC​GTC​AT 3’

sgRNA2: 5′ TAA​CAG​ATC​AAC​CTC​ACC​TC 3’

ssODN repair template (sequence of the lq190 mutation): 5′GAA​AAT​GCT​TTA​TCT​GGC​TGT​TTC​GGA​CAC​CTT​AACCGA​CC​AGC​CAC​ATC​TTA​AGC​CAA​TTG​AGT​CTG​CCC​TTT​CTCCT​G​TTA​TTG​TCAT​CGG​ACA​ATG​TGA​TGT​TTC​AAT​GTC​AGC​TC​A​TG3’ (recoded exon sequence is underlined).

Proportional data were analyzed using Fisher’s Exact test. Continuous data were analyzed with Student’s t-test. Sample sizes used each gave an estimated statistical power of >80% (a β false negative value of <0.2), with an ⍺ false positive value of 0.05.

VD/DD axons were visualized with Punc-25::gfp transgene, juIs76 (JIN et al. 1999), which is expressed in 13 VD and 6DD GABAergic motor neurons. Axon guidance defects were scored as previously described (Mahadik and Lundquist, 2022). An average of 16 of the 19 commissures of VD/DD axons are distinguishable in wild-type, due to fasciculation and inability to distinguish individual axons. A total of 100 animals were scored (1,600 total commissural processes). At a false positive a value of 0.05 with observed variances of 5% or greater, scoring 1,600 axons gives an estimated statistical power of ≥80%. Total axon guidance defects were calculated by counting all the axons which failed to reach the dorsal nerve cord, wandered at an angle of 45^° or greater, crossed over other processes, and displayed ectopic branching. Significance difference between two genotypes was determined by using Fisher’s exact test.

Growth cones of VD axons were imaged as previously described (Norris and Lundquist, 2011; Norris et al., 2014; Gujar et al., 2017; Gujar et al., 2018; Gujar et al., 2019; Mahadik and Lundquist, 2022; Mahadik and Lundquist, 2023). Late L1/early L2 larval animals placed on a 2% agarose pad with 5 mM sodium azide in M9 buffer. At n = 50, at a false positive a value of 0.05, statistical power of β (false negative) is much lower than 0.2. Growth cones were imaged with Qimaging Retiga EXi camera on a Leica DM5500 microscope at ×100 magnification. Projections less than 0.5 um in width were scored as filopodia. Growth cone area and filopodial length were quantified using ImageJ software as previously described (Norris and Lundquist, 2011; Norris et al., 2014; Gujar et al., 2017; Gujar et al., 2018; Gujar et al., 2019; Mahadik and Lundquist, 2022; Mahadik and Lundquist, 2023). Significance of difference between two genotypes was determined by two-sided t-test with unequal variance.

Polarity of growth filopodial protrusions was determined as previously described (Norris and Lundquist, 2011; Norris et al., 2014; Gujar et al., 2017; Gujar et al., 2018; Gujar et al., 2019; Mahadik and Lundquist, 2022; Mahadik and Lundquist, 2023). Growth cone images were divided into dorsal and ventral halves with respect to the ventral nerve cord. The number of filopodia in each was counted. Proportion of dorsal filipodia was determined by the number of dorsal filopodia divided by total number of filopodia. Significance of difference between two genotypes was determined by Fisher’s exact test.

Growth cone F-actin was analyzed as previously described (Gujar et al., 2018), using the F-actin binding domain of VAB-10/spectraplakin fused to GFP (Bosher et al., 2003; Patel et al., 2008). A soluble mCherry volume marker was included in the strain. Growth cones images were captured as described above. ImageJ was used image analysis to determine asymmetric VAB-10ABD::GFP localization. The pixel intensities for VAB-10ABD::GFP were normalized to the volumetric mCherry fluorescence in five line scans from the dorsal half and the ventral half of each growth cone. This normalized ratio was determined for multiple growth cones, and the average and standard error for multiple growth cones was determined. Statistical comparisons between genotypes were done using a two-tailed t-test with unequal variance on these average normalized ratios of multiple growth cones of each genotype.

Growth cone EBP-2::GFP puncta (Srayko et al., 2005; Kozlowski et al., 2007; Yan et al., 2013) was analyzed as described previously ( Gujar et al., 2018). Growth cone perimeter and filopodia were defined, and the EBP-2:GFP puncta in the growth cone were counted. Significance of difference was determined by a two-sided t-test with unequal variance.