Netrin/UNC-6 triggers actin assembly and non-muscle myosin activity to drive dendrite retraction in the self-avoidance response

Dendrite growth is constrained by the self-avoidance response but the downstream pathways that balance these opposing mechanisms are unknown. We have proposed that the diffusible cue UNC-6(Netrin) is captured by UNC-40 (DCC) for a short-range interaction with UNC-5 to trigger self-avoidance in the C. elegans PVD neuron. Here we report that the actin-polymerizing proteins UNC-34(Ena/VASP), WSP-1(WASP), UNC-73(Trio), MIG-10(Lamellipodin) and the Arp2/3 complex effect dendrite retraction in the self-avoidance response mediated by UNC-6(Netrin). The paradoxical idea that actin polymerization results in shorter rather than longer dendrites is explained by our finding that NMY-1 (non-muscle myosin II) is necessary for retraction and could therefore mediate this effect in a contractile mechanism. Our results also show that dendrite length is determined by the antagonistic effects on the actin cytoskeleton of separate sets of effectors for retraction mediated by UNC-6(Netrin) versus outgrowth promoted by the DMA-1 receptor. Thus, our findings suggest that the dendrite length depends on an intrinsic mechanism that balances distinct modes of actin assembly for growth versus retraction.

receptor on an opposing sister dendrite to induce retraction. Here we report genetic evidence that 48 UNC-5-dependent retraction requires downstream actin polymerization. This finding evokes a 49 paradox: How might actin polymerization drive both dendrite growth and retraction? We propose 50 two answers: (1) Distinct sets of effectors are involved in actin assembly for growth vs retraction; 51 (2) Non-muscle myosin interacts with a nascent actin assemblage to trigger retraction. Our results 52 show that dendrite length depends on the balanced effects of specific molecular components that 53 induce growth vs retraction. 54 55 56 57 58 59 60 61 All rights reserved. No reuse allowed without permission.

INTRODUCTION
of unc-34/Ena/VASP mutants which our results suggest are due to a non-cell autonomous function 242 for UNC-34 ( Figure S2). 243 To test for potential roles for branched actin polymerization, we examined a null allele of wsp-244 1/WASP (Wiskott-Aldrich Syndrome protein) and observed defective PVD self-avoidance. If wsp-245 1/WASP functions in a common pathway with unc-6, then the self-avoidance defect of the double 246 mutant, wsp-1; unc-6, should be no more severe than that of either wsp-1 or unc-6 single mutants. 247 We confirmed this prediction by scoring PVD self-avoidance in wsp-1; unc-6 mutant animals 248 ( Figure 3D). An independent study confirmed the role of wsp-1 in PVD self-avoidance. This work 249 also reported, however, that wsp-1 does not enhance the self-avoidance defect of a parallel-acting for dendrite retraction and that this mechanism is activated by UNC-6/Netrin in the self-avoidance 262 response. 263 264 All rights reserved. No reuse allowed without permission.

13
The actin cytoskeleton is highly dynamic in growing and retracting PVD dendrites. 265 To test the idea that the actin cytoskeleton is differentially modulated for both outgrowth and 266 retraction we expressed LifeAct::GFP in PVD neurons to monitor actin dynamics during dendrite 267 morphogenesis. Although LifeAct::GFP binds both filamentous actin (F-actin) and monomeric G-268 actin with high affinity, bright LifeAct::GFP-labeled foci typically correspond to F-actin-269 containing structures [39]. We observed that the LifeAct::GFP signal was consistently brightest that F-actin is required, we monitored the LifeACT::GFP signal at the tips of 3° dendrites during 282 the self-avoidance response ( Figure S4) (Movie S3). The cytoplasmic mCherry marker was used 283 to detect contact events as instances in which the measured gap between left and right 3° dendrites 284 approached zero. Although time-lapse imaging detected dynamic fluctuations in LifeAct::GFP 285 fluorescence both before and after contact, we did not observe a quantitative increase in the 286 LifeACT::GFP signal with retraction ( Figure S4 and S5). Notably, for these experiments, we used 287 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/492769 doi: bioRxiv preprint the marker strains (i.e., PVD::LifeAct::GFP and PVD::mcherry), imaging strategy (i.e., 40 X 288 objective, 35 s sampling interval) and method of quantitative analysis described in an earlier 289 publication that reported increased LifeAct::GFP fluorescence after contact (N = 14, p = 0.54, two-290 way ANOVA with Bonferroni correction [37] ( Figure S4A-C). In an additional approach, we used 291 a higher resolution objective (100X) and longer sampling interval (1 -1.3 min) and also did not 292 detect elevated LifeAct::GFP fluorescence during retraction (N = 8, p = 0.74, two-way ANOVA 293 with Bonferroni correction) ( Figure S4D and S5). The simplest explanation of our results is that 294 actin polymerization drives both outgrowth and retraction which are tightly coupled over time at 295 dendrite tips and thus difficult to resolve as separate events. 296 297 NMY-1/Non-muscle myosin II drives dendrite retraction in the self-avoidance mechanism. 298 Our imaging results indicate that F-actin is the dominant cytoskeletal structure in growing PVD 299 dendrites. In considering a mechanism to account for dendrite shortening in the self-avoidance 300 response, we hypothesized that a newly synthesized actin cytoskeleton could mediate dendrite 301 shortening by providing a substrate for non-muscle myosin II to drive contraction. To test this 302 idea, we examined loss-of-function alleles that disable the C. elegans non-muscle myosins, NMY-303 1 and NMY-2. This experiment determined that mutations in nmy-1 but not nmy-2 [40] show PVD 304 self-avoidance defects ( Figure 6). This effect is cell autonomous for nmy-1 because PVD-specific 305 RNAi of nmy-1 also results in overlapping 3° dendrites and PVD-specific expression of GFP-306 tagged NMY-1 is sufficient to rescue the nmy-1 self-avoidance defect (Figure 6 B-D). We conclude 307 that a specific non-muscle myosin, NMY-1, promotes dendrite retraction in the self-avoidance 308 mechanism and could act by driving the translocation of F-actin bundles at the tips of 3° dendrites. 309 310 All rights reserved. No reuse allowed without permission.

Phosphomimetic activation of MLC-4 shortens PVD 3° dendrites. 311
Non-muscle myosin II motor proteins are composed of myosin heavy chain (MHC) and light 312 chains, the essential light chain (ELC) and regulatory light chain (RLC) [41,42]. Myosin motor 313 activity is triggered by RLC phosphorylation that induces a conformational change to allow 314 assembly of the myosin complex into active bipolar filaments ( Figure 7A) [43]. Having shown 315 that NMY-1/non-muscle myosin is required for self-avoidance, we next performed an additional 316 experiment to ask if constitutive non-muscle myosin activity is sufficient to induce dendrite 317 retraction. For this test, we expressed a phosphomimetic mutant of MLC-4, the C. elegans RLC 318 homolog. MLC-4/myosin regulatory light chain contains serine/S and threonine/T phosphorylation 319 sites in a highly-conserved domain. Both S and T residues were mutated to Aspartate/D and the 320 resultant phosphomimetic construct MLC-4DD was fused to GFP for transgenic expression in 321 PVD ( Figure 7B-C) [44]. PVD neurons that expressed MLC-4DD showed significantly shorter 3° 322 dendrites at the L3 stage in comparison to wild type as well as a concurrent increase in the width 323 of gaps between 3° dendrites from adjacent menorahs ( Figure 7D-G). These phenotypic traits are 324 consistent with the idea that normal dendrite outgrowth is slowed by constitutive activation of non-325 muscle myosin by MLC-4ADD. By the L4 stage, PVD architecture in the MLC-4DD strain 326 resembles that of the wild type with normal appearing menorah morphology and spacing ( Figure  327 8A, B). Similar results were observed for PVD neurons that express constitutively active 328 MYR::UNC-5 which we have proposed functions downstream of UNC-6/Netrin to trigger the self-329 avoidance response (Figure 2). Our finding that MLC-4DD rescues the self-avoidance defect of 330 unc-6 mutants suggests that NMY-1/non-muscle myosin II also functions downstream of unc-6 331 for dendrite retraction ( Figure 8C). 332 333 All rights reserved. No reuse allowed without permission.

Antagonistic pathways regulate dendrite outgrowth vs self-avoidance. 334
Our time lapse imaging studies detected dynamic actin polymerization in growing PVD dendrites 335 (Movie S3). This observation and the recent finding that PVD dendrites are truncated and 336 misdirected by mutations that disable either a specific actin structural gene, act-4, or specific 337 components (WRC and TIAM/GEF) that promote actin polymerization, suggest that dendritic 338 outgrowth depends on actin assembly [26,27] ( Figure S2D). Actin assembly is also likely required 339 for dendrite retraction given results showing that genetic knockdown of a separate set of actin-340 binding proteins (i.e., UNC-34/Ena/VASP, MIG-10/Lpd, UNC-73/Trio, WSP-1/WASP, ARX-341 5/p21) (Fig. 3), blocks PVD self-avoidance but not dendrite outgrowth [37]. A shared role for 342 actin polymerization in both outgrowth and retraction argues that the net length of PVD dendrites 343 must depend on an intrinsic mechanism that regulates actin assembly to balance these competing 344 effects. 345

346
Our results suggest that UNC-6/Netrin signaling promotes dendrite retraction in a mechanism 347 involving F-actin assembly. PVD dendritic outgrowth is driven by a separate pathway in which  kpc-1 activity is proposed to temporally weaken adhesion with the epidermis to facilitate changes 354 in trajectory at specific stages in dendritic outgrowth [47,48]. For example, the tips of 3° branches 355 typically execute a right-angle turn after the self-avoidance response to produce a 4° dendrite [11]. 356 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. Thus, we considered the idea that self-avoidance is achieved by the combined effects of kpc-1-359 dependent downregulation of DMA-1 and a separate mechanism that antagonizes outgrowth and 360 requires UNC-6/Netrin. We performed a series of genetic experiments to test this model. 361
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/492769 doi: bioRxiv preprint ( Figure 9C, D). Thus, these results suggest that DMA-1-mediated adhesion to the epidermis 380 antagonizes UNC-6/Netrin-dependent dendrite retraction and that the relative strengths of these 381 opposing pathways must be finely tuned to achieve the self-avoidance effect ( Figure 9E). The key 382 role of actin polymerization in both UNC-6/Netrin-mediated retraction and DMA-1-dependent 383 outgrowth argues that this balancing mechanism likely regulates both modes of actin assembly. 384 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. Our finding that non-muscle myosin is also involved resolves the paradoxical idea that UNC-401 6/Netrin-dependent actin polymerization shortens rather than lengthens PVD 3° dendrites. We 402 propose that non-muscle myosin engages the F-actin cytoskeleton at the tips of PVD dendrites to 403 drive retraction. 404 405

Regulators of actin polymerization are required for dendrite self-avoidance. 406
All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/492769 doi: bioRxiv preprint Our genetic analysis detected a cell-autonomous role for NMY-1/non-muscle myosin II in dendrite 452 self-avoidance ( Figure 6). Because our results revealed F-actin at the tips of PVD dendrites ( Figure  453 5), we suggest that NMY-1 mediates retraction by interacting with F-actin to drive a contractile 454 mechanism. For example, NMY-1 could accelerate retrograde flow, a non-muscle myosin-455 dependent effect that also drives filopodial retraction in growth cones and at the leading edge of Thus, our findings suggest that key drivers of axonal repulsion including nascent actin assembly 471 and non-muscle myosin may also mediate dendrite retraction in the self-avoidance response. 472 473 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/492769 doi: bioRxiv preprint RLC phosphorylation activates non-muscle myosin II by releasing myosin monomers for assembly 474 into bipolar structures or myosin "stacks" that interact with actin filaments to induce translocation 475 Although self-avoidance is perturbed in UNC-6/Netrin pathway mutants, a significant fraction 490 (~70%) ( Figure 3D) of PVD 3° dendrites show apparently normal self-avoidance behavior [9]. 491 One explanation for this observation is provided by the recent discovery of an independent 492 pathway, mediated by MIG/14/Wntless, that is necessary for self-avoidance in an additional 493 fraction of PVD 3° dendrites [37]. Parallel acting pathways also mediate self-avoidance in 494 mammalian Purkinje neurons where slit-robo signaling acts in concert with protocadherins [6,8]. 495 The likelihood of additional effectors of self-avoidance is also suggested by the incompletely 496 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/492769 doi: bioRxiv preprint effects is readily observed in the development of 3° PVD dendrites. Initially, 3° dendrites 520 emanating from adjacent menorahs grow out along the body axis until contacting one another and 521 then retract to avoid overlap [9,11]. Outgrowth in this case is promoted by a multicomponent 522 receptor-ligand complex that mediates adhesive interactions with the adjacent epidermis [70-73] 523 ( Figure 9E). When this pathway is dysregulated by over-expression of the DMA-1 receptor in 524 PVD neurons, for example, 3° dendrites continue to adhere to the epidermis and overgrow one 525 another [47,74]. 3° branch self-avoidance also fails in mutants that disable UNC-6/Netrin signaling 526 which we have shown promotes dendrite retraction [9]. Thus, these results suggest that the net 527 length and placement of each 3° branch depends on the balanced effects of locally acting signals 528 that either extend (e.g., DMA-1) or shorten (e.g., UNC-6/Netrin) the dendrite. We confirmed this 529 idea in genetic experiments that detected strong dose-sensitive interactions between these opposing 530 pathways. For example, a 50% reduction of UNC-6/Netrin expression in a heterozygous unc-6/+ 531 mutant does not perturb self-avoidance. Similarly, a genetic mutant (kpc-1/+) that partially 532 elevates DMA-1 also shows normal 3° branch outgrowth. The combination of both mutations in 533 the double heterozygote, unc-6/+; kpc-1/+, however, produces a strong self-avoidance defect 534 ( Figure 9). Our genetic results identified multiple regulators of actin polymerization (UNC-535 34/Ena/VASP, WSP-1/WASP, ARX-5/p21, MIG-10/Lamelipodin, UNC-73/Trio) that are 536 required for self-avoidance (Figure 3). We thus propose that UNC-6/Netrin promotes actin 537 polymerization to effect dendrite retraction and that non-muscle myosin II interacts with a nascent 538 actin cytoskeleton to translocate each 3° dendrite away from its neighbor. F-actin is also abundant 539 and highly dynamic in growing PVD dendrites. Notably, a distinct set of F-actin promoting factors 540 (WRC, TIAM/GEF) interacts with the DMA-1 receptor complex to drive PVD dendritic growth 541 ( Figure 9E) [26,27]. Additional genetic evidence suggests that the WRC could also function in 542 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/492769 doi: bioRxiv preprint self-avoidance [37]. Thus, the challenge for future studies is to elucidate the cell biological 543 machinery through which opposing pathways manipulate the actin cytoskeleton to effect either 544 dendrite growth or retraction. 545 546 All rights reserved. No reuse allowed without permission.

Strains and genetics 549
All the strains used were maintained at 20°C and cultured as previously described (Brenner, 1974). 550 We used the N2 Bristol strain as wild-type. corresponding to pCJS95 (pF49H12.4::GFP::CED-10::unc-10 3'UTR). We also PCR amplified 586 pCJS95 with primers overlapping the nmy-1 genomic sequences to swap CED-10 with NMY-1. 587 We then used the NEB site-directed mutagenesis kit to insert a 24-nucleotide glycine rich linker 588 (see Table 1) [65]. The resultant pLSR11 plasmid was sequenced to confirm that the linker was 589 inserted between the GFP sequence and nmy-1. See table 1 for primer sequences. 590 591 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. were assembled using image J. Z stack images that were collected using the LAS-AF or NIS 609 elements software was imported into ImageJ/Fiji for further analysis. retractions during the length of the time lapse videos were analyzed. NIS elements was used for 636 measurements of LifeAct::GFP from a 1µm ROI at the tips of growing dendrites. In the same 637 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/492769 doi: bioRxiv preprint videos, fluorescence intensity was measured for an identical 1µm ROI of PVD branches (1º, 2º 638 and 3º) that were not undergoing outgrowth. Fluorescence intensity measurements for the 639 outgrowth and no-outgrowth datasets were normalized against the maximum overall LifeAct::GFP 640 value ( Figure 5). 641 642 Statistics: Student's t test was used for comparisons between 2 groups and ANOVA for 643 comparisons between three or more groups with either Tukey posthoc or Bonferroni correction for 644 multiple comparisons. Kolmogorov-Smirnov test was performed for comparisons between 645 frequency distributions of 3° dendrite length and gaps between 3° dendrites ( Figures 2C and 8F  The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.  avoidance defects (percentage of overlapping 3° branches) are rare (< 5%) in wild type and 933 PVD::MYR::UNC-5. The Unc-6 self-avoidance defect is prevented by PVD expression of 934 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. Merge 0.00 min 1.00 min 2.00 min 3.00 min 12.00 min All rights reserved. No reuse allowed without permission.

C.
A. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.