DAPK interacts with Patronin and the microtubule cytoskeleton in epidermal development and wound repair

Epidermal barrier epithelia form a first line of defense against the environment, protecting animals against infection and repairing physical damage. In C. elegans, death-associated protein kinase (DAPK-1) regulates epidermal morphogenesis, innate immunity and wound repair. Combining genetic suppressor screens and pharmacological tests, we find that DAPK-1 maintains epidermal tissue integrity through regulation of the microtubule (MT) cytoskeleton. dapk-1 epidermal phenotypes are suppressed by treatment with microtubule-destabilizing drugs and mimicked or enhanced by microtubule-stabilizing drugs. Loss of function in ptrn-1, the C. elegans member of the Patronin/Nezha/CAMSAP family of MT minus-end binding proteins, suppresses dapk-1 epidermal and innate immunity phenotypes. Over-expression of the MT-binding CKK domain of PTRN-1 triggers epidermal and immunity defects resembling those of dapk-1 mutants, and PTRN-1 localization is regulated by DAPK-1. DAPK-1 and PTRN-1 physically interact in co-immunoprecipitation experiments, and DAPK-1 itself undergoes MT-dependent transport. Our results uncover an unexpected interdependence of DAPK-1 and the microtubule cytoskeleton in maintenance of epidermal integrity. DOI: http://dx.doi.org/10.7554/eLife.15833.001


Introduction
Death-associated protein kinase 1 (DAPK1) and its related calcium-regulated serine/threonine kinases play a wide variety of roles in cell death and tumor suppression (Bialik and Kimchi, 2006). Mammalian DAPK1 has been implicated in stress responses (Tu et al., 2010), antiviral immunity (Zhang et al.), and in IL-1ß-associated inflammatory diseases (Chakilam et al., 2013;Chuang et al., 2011). In addition, DAPK1 can act as a checkpoint in the macrophage inflammation program (Mukhopadhyay et al., 2008) and as a negative regulator of T-cell receptor-mediated activation of NFkB (Chuang et al., 2008).
In the nematode C. elegans the sole DAPK family member, DAPK-1, plays roles in autophagy and in excitotoxic neuronal death (Del Rosario et al., 2015;Kang and Avery, 2010). DAPK-1 also regulates epidermal development and wound repair, independently of known cell death programs. The nematode epidermis is a barrier epithelium that forms the first line of defense against environmental stresses, such as pathogens and physical damage (Engelmann and Pujol, 2010). Mutations in dapk-1 result in progressive degeneration of the epidermis, cuticle hypertrophy, and constitutive activation of epidermal innate immune responses via a p38 MAPK cascade (Tong et al., 2009). Such dapk-1 mutants behave as if they are constitutively wounded even without injury, and when wounded exhibit faster wound repair (Xu and Chisholm, 2011). How DAPK-1 regulates these multiple aspects of epidermal maintenance and wound repair is not yet understood.
Here we took a genetic approach to understanding DAPK-1's functions in the epidermis. We identified genetic suppressors and enhancers of dapk-1 morphological defects, revealing novel roles for microtubule (MT) regulators in the mature epidermis. Genetic and pharmacological manipulations suggest aberrant dapk-1 function causes excessive MT stabilization, resulting in morphological defects. dapk-1 epidermal defects can be suppressed by loss of function in the MT minus end binding protein PTRN-1 and by pharmacological destabilization of MTs. Moreover, overexpression of the MT-binding domain of PTRN-1 is sufficient to induce dapk-1-like epidermal defects. Our data suggest DAPK-1 destabilizes epidermal MTs by inhibiting the function of PTRN-1. We further show for the first time that DAPK-1 itself undergoes MT-dependent transport. Our findings reveal an unexpected interplay between DAPK-1, the epithelial MT cytoskeleton, and epidermal morphology and wound repair.

dapk-1 epidermal morphological defects can be suppressed or enhanced by loss of function in microtubule regulators
To identify new dapk-1 interactors, we screened for genetic suppressors of dapk-1 epidermal phenotypes. All dapk-1 mutants display epidermal morphological (Mor) defects, with penetrance varying depending on the allele ( Figure 1A, Figure 1-figure supplement 1A) (Tong et al., 2009). The Mor phenotype reflects a progressive accumulation of the cuticle and degeneration of the underlying epidermis at the extreme anterior and posterior, as well as the dorsal midline. dapk-1(ju4), which causes a missense alteration S179L in the DAPK-1 kinase domain, causes 100% of animals to display this aberrant morphology. We mutagenized dapk-1(ju4) animals and screened for suppression of the Mor phenotype (see Methods). We identified multiple extragenic suppressors of dapk-1(ju4), two of which are described here. One suppressor, ju698, causes a nonsense mutation in ptrn-1, which encodes the C. elegans member of the Patronin/CAMSAP/Nezha family of MT minus end binding proteins. A null allele ptrn-1(lt1) suppressed dapk-1(ju4) phenotypes to the same extent as ptrn-1(ju698) ( Figure 1B; Figure 1-figure supplement 1C). Suppression of dapk-1(ju4) by ptrn-1(0) was rescued by a single-copy insertion (Mos-SCI) ptrn-1(+) transgene and by transgenes expressing PTRN-1 under the control of the dpy-7 promoter, specific to the larval epidermis, indicating that loss of PTRN-1 function in the larval epidermis is required for suppression of the dapk-1(ju4) phenotype.
Based on our identification of two MT-interacting proteins in our suppressor screen we tested additional MT-associated factors, as well as orthologs of genes known to interact with DAPK1 or CAMSAPs (Table 1). Partial loss of function in unc-116, which encodes a kinesin-1 plus-end directed motor, suppressed dapk-1(ju4) morphological defects. DAPK family members have previously been implicated in MT stability, promoting function of the MT-associated protein tau via the Pin1 prolyl isomerase or the MARK kinase Wu et al., 2011). However, loss of function in orthologs of these genes (ptl-1, pinn-1, par-1) did not modify dapk-1(ju4) phenotypes (Table 1), suggesting DAPK-1 regulates epidermal MTs via a novel mechanism.
CAMSAPs are thought to promote MT stability in part by protecting minus ends from the MT depolymerizing enzyme kinesin-13 (Goodwin and Vale, 2010). Consistent with this model, loss of function in klp-7/kinesin-13 strongly enhanced morphological defects of dapk-1(gk219), both in ptrn-1(+) and in ptrn-1(0) backgrounds ( Figure 1B). Partial loss of function of the MT severing enzymes MEI-1/p60 katanin or SPAS-1/Spastin also enhanced dapk-1(gk219) Mor phenotypes. None of these mutants conferred morphological defects in a dapk-1(+) background ( Figure 1 T d a p k -1 ( j u 4 ) p t r n -1 ( j u 6 9 8 ) p t r n -1 ( l t 1 ) l t 1 [ P T R  The following figure supplement is available for figure 1:  1D). These analyses suggest that aberrant dapk-1 function causes epidermal integrity to be sensitized to MT stability. dapk-1(ju4) defects were not suppressed by loss of function in several other genes implicated in MT dynamics ( Table 1), suggesting a specific subset of MT regulators can affect epidermal morphogenesis.
dapk-1 mutants also display accelerated epidermal wound closure, manifested by the rate of closure of actin rings that form around puncture wounds (Xu and Chisholm, 2011). We found that ptrn-1 mutations suppressed this accelerated wound closure to wild type rates ( Figure 2C Pharmacological modulation of MT stability can suppress or enhance dapk-1 morphological defects Because the dapk-1 epidermal phenotypes are suppressed by loss of function in a MT stabilizing factor (Patronin/ptrn-1) and enhanced by loss of function in MT destabilizing factors (Kinesin-13/klp-7, Katanin/mei-1, Spastin/spas-1) we hypothesized that epidermal defects in dapk-1 mutants might result from excessive stabilization of epidermal MTs. We therefore tested whether drugs that depolymerize MTs (colchicine, nocodazole) or stabilize MTs (paclitaxel) could modify epidermal Mor defects. Some experiments used cat-4 mutants, which are defective in biopterin synthesis and display leaky cuticles and hypersensitivity to drugs (Loer et al., 2015).
ptrn-1(0) mutants display overtly normal epidermal morphology, but were six times more sensitive to the effects of paclitaxel compared to WT animals ( Figure 3A,B). Paclitaxel also induced morphological defects in dapk-1(gk219) ptrn-1(0) double mutants (Figure 3-figure supplement 1A), although not to the same extent as in dapk-1(gk219) single mutants ( Figure 3A). dhc-1(or195) single mutants were also hypersensitive to paclitaxel, compared to wild type (Figure 3-figure supplement 1A), possibly reflecting increased free tubulin concentration in these mutants. Taken together, these data are consistent with the model that dapk-1 mutants display excessively stabilized MTs that cause aberrant epidermal morphology.

dapk-1 mutants display aberrant MTs resembling those of paclitaxelstabilized animals
The above analyses suggest dapk-1 mutants might display excessively stabilized MTs. We therefore analyzed epidermal MT architecture using the tubulin marker TBB-2::GFP (see Materials and methods; Figure 4-figure supplement 1A). In the lateral hyp7 epidermal syncytium MTs form a dense meshwork mostly oriented along the anteroposterior axis. In contrast, the dorsoventral compartment of hyp7, overlying body wall muscles, contains parallel circumferential MT bundles spaced 1-1.5 mm apart ( Figure 3C; Figure 4-figure supplement 1B,C; Figure 4E). These arrays extend from the lateral epidermis to the dorsal or ventral midlines.
Chronic treatment with colchicine caused breakage and loss of circumferential MT bundles in the dorsoventral epidermis ( Figure 3D); in the lateral compartment, MTs became sparse and less bundled than in the WT. Conversely, paclitaxel treatment caused circumferential MT bundles to be straighter and thicker ( Figure 3D,E). At high paclitaxel concentrations (15 mM), circumferential MT bundles became disorganized and lateral MT bundles were thicker ( Figure 3D,E). Thus, MT stabilization by paclitaxel results in MTs becoming straighter and more bundled.
We then examined MTs in dapk-1(ju4) mutants, focusing on the anterior epidermis, where dapk-1 morphological defects begin. In early (L2) larvae the MT architecture of dapk-1(ju4) animals appeared normal ( Figure 3C; Figure 4-figure supplement 1B,C). By the late larval (L4) stage a region devoid of MTs appeared in the anterior lateral epidermis of dapk-1(ju4) animals ( Figure 3C, asterisk), apparently due to local degeneration of the epidermis and cuticle overgrowth (Tong et al., 2009). Nearby circumferential MT bundles were disorganized ( Figure 3C In the lateral epidermis of dapk-1(ju4) animals MT bundles were straighter and more bundled than in wild type ( Figure 4A, yellow arrowheads), resembling those of paclitaxeltreated animals. We used FibrilTool to measure the anisotropy (directional dependence) of lateral MT bundles (see Methods) and found that dapk-1(ju4) mutants displayed increased MT bundle anisotropy compared to wild type ( Figure 4C). The MT defects of dapk-1(gk219) mutants were very similar to those of dapk-1(ju4) (Figure 4-figure supplement 1D-H). The similarities between the dapk-1 mutant phenotypes and the effects of paclitaxel treatment suggest DAPK-1 normally destabilizes epidermal MTs.
We next assessed the effects of dapk-1 mutants on epidermal MT dynamics using the MT plus end marker EBP-2::GFP (EBP-GFP for brevity). EBP-GFP binds to growing MT plus ends and in vivo forms moving comets whose movement can be quantified by kymograph analysis (see Methods). We found that MT plus ends in the wild type adult epidermis were highly dynamic, with 0.09 ± 0.015 comets/mm 2 in the lateral epidermis, and slightly more (0.12 ± 0.015 comets/mm 2 ) in the dorsoventral epidermis (Videos 1,2). The overall density of comets in ptrn-1(0) animals was similar to that in wild type ( Figure 4H), as previously reported for larvae . By contrast, dapk-1(ju4) mutants, as well as dapk-1(ju4) ptrn-1(0) double mutants, had significantly more comets in the lateral epidermis than in WT or ptrn-1(0) (Videos 3,4). In wild type adults, EBP-GFP comets grew at 0. . Left column, lateral views of the head; right column, dorsal view. ptrn-1(0) mutant adults display fewer circumferential MT bundles in the dorsoventral epidermis (red asterisks). dapk-1(ju4) mutants display increased bundling in the lateral epidermis (yellow arrowhead), quantified by anisotropy index in panel C. dapk-1(ju4) ptrn-1(0) double mutants display normal MT bundling in Figure 4 continued on next page 0.006 mm/s in the lateral epidermis, and 0.26 ± 0.01 mm/s in the dorsoventral epidermis, comparable to growth rates in the larval epidermis . MT growth rates in the lateral epidermis were significantly reduced in dapk-1(ju4) mutants, and increased in ptrn-1(0) single and in dapk-1 (ju4) ptrn-1(0) mutants ( Figure 4G). dapk-1(ju4) mutants thus display more slow-growing MT plus ends in the lateral epidermis, consistent with a partial stabilization of MT dynamics.
Finally, we analyzed the directionality of plus-end growth. In the lateral epidermis most EBP-GFP comets grew anteriorly or posteriorly, without a strong bias in directionality. Similarly, in the dorsoventral epidermis equal numbers of EBP-GFP comets grew towards or away from lateral epidermal ridges ( Figure 4I). EBP-GFP comet directionality was normal in ptrn-1(0) mutant and dapk-1(ju4) ptrn-1(0) backgrounds. However, dapk-1(ju4) mutants displayed a significant bias in comet directionality in the dorsoventral epidermis, such that fewer comets grew away from the lateral epidermis. Thus, in dapk-1 mutants growing MTs are more confined to the lateral epidermis and less likely to extend into the dorsoventral epidermis.

Over-expression of the PTRN-1 CKK domain induces dapk-1-like morphology defects
To understand how PTRN-1 might regulate epidermal MTs in the dapk-1 mutant we tested individual PTRN-1 domains. Like other CAMSAP proteins, PTRN-1 has three conserved regions: an N-terminal calponin homology (CH) domain, of unknown function; a central coiled-coil (CC) domain, known to interact with cytoskeleton associated proteins, and a C-terminal MT-binding CKK domain specific to CAMSAPs ( Figure 5A). We expressed GFP-tagged fragments of PTRN-1 in the larval epidermis of dapk-1(ju4) ptrn-1(0) mutants as multicopy transgenes using the dpy-7 promoter. As shown above, expression of full-length PTRN-1 rescued the ptrn-1(0) suppression phenotypes ( Figure 1B). Expression of the CKK domain alone also restored the dapk-1 epidermal morphology phenotype in dapk-1 (ju4) ptrn-1(0) double mutants, whereas constructs lacking the CKK domain could not rescue, suggesting the CKK domain is required for PTRN-1 function ( Figure 5B). However, constructs containing the CKK domain and either the CC or CH domain (i.e. DCH or DCC respectively) had significantly weaker rescuing activity compared to the CKK domain alone ( Figure 5B). These observations suggest that the CKK domain is critical for PTRN-1 function in the epidermis and is inhibited by the CH or CC domains.
Strikingly, expression of the CKK domain alone in ptrn-1(0) mutants caused highly penetrant Mor phenotypes, resembling those of dapk-1(ju4) (Figure 5B,C). This is in contrast to transgenic animals expressing full-length PTRN-1, which do not display such phenotypes. The CKK domain acts cell autonomously in the epidermis, as a pan-neuronal expression of the CKK domain in ptrn-1(0) mutants did not induce Mor phenotypes (not shown). Paralleling our rescue analysis, transgenes expressing the CKK domain and the CH or CC domains did not cause epidermal defects in ptrn-1(0) animals. Moreover, expression of the CKK domain only caused aberrant development in a ptrn-1(0) background and not in a wild type background ( Figure 5B), suggesting that CKK domain activity is inhibited by endogenous PTRN-1. In addition, expression of the CKK domain in ptrn-1(0) mutants induced Pnlp-29-GFP expression, similar to that seen in dapk-1 (ju4) mutants ( Figure 5D), and accelerated wound closure ( Figure 5E). These data are consistent with DAPK-1 and PTRN-1 acting antagonistically in epidermal development, and suggest that DAPK-1 might specifically inhibit the activity of the PTRN-1 CKK domain.

Localization of PTRN-1 along MTs, mediated by its CKK domain, correlates with defective epidermal morphogenesis
In other CAMSAP proteins the CKK domain binds MTs (Baines et al., 2009), whereas the CC domain confers minus-end targeting (Goodwin and Vale, 2010). We therefore investigated whether PTRN-1 localization correlated with its effects on epidermal morphology. Full-length GFP::PTRN-1 localized to puncta and to short filaments that were either thin (0.22 ± 0.01 mm wide) or thick (0.38 ± 0.01 mm) ( Figure 5F); the latter co-localized with MTs ( Figure 5-figure supplement 1E). In contrast, the CKK domain localized to longer thin filaments ( Figure 5G; Figure 5-figure supplement 1C) and colocalized with MTs ( Figure 5H). GFP::PTRN-1(DCKK) was almost completely punctate ( Figure 5G). Taken together, the CKK domain is critical for localization along MTs, and MT localization is necessary but insufficient to trigger aberrant epidermal morphology ( Table 2).
The PTRN-1 CH domain did not confer subcellular localization, whereas fragments lacking the CH domain localized to filaments and puncta, resembling full length PTRN-1 ( Figure 5G, Table 2). The PTRN-1 CC domain localized primarily to puncta, whereas constructs lacking this domain (DCC) did not form puncta or thick filaments ( Figure 5G). Further dissections suggest the coiled-coil subregions of the CC domain have distinct roles such that CC1 and CC2 promote localization to puncta and thick filaments ( Figure 5A We hypothesized that the expression of the CKK domain in the absence of other PTRN-1 domains causes excessive MT stabilization and aberrant epidermal development. Both fulllength PTRN-1 and CKK domain transgenes restored circumferential MTs to ptrn-1(0) mutants ( Figure 5-figure supplement 1D). However, expression of the CKK domain in a ptrn-1(0) background caused circumferential MT bundles to be straighter than in wild type, resembling the effects of paclitaxel treatment (Figure 4-figure supplement 1D). ptrn-1(0) mutants expressing full-length PTRN-1 did not display straighter MT bundles or Mor phenotypes ( Figure 5-figure supplement 1D). Thus the MT-stabilizing effects of the CKK domain can induce aberrant epidermal development, and are normally inhibited by other PTRN-1 domains.

DAPK-1 undergoes directional MT-dependent transport in the epidermis
Our genetic and pharmacological analysis of the dapk-1 mutant defect in epidermal morphology suggests DAPK-1 might inhibit MT stability, possibly via negative regulation of PTRN-1 itself. DAPK-1 might act directly or indirectly on PTRN-1. Although DAPK family members have been shown to localize to the cytoskeleton, the subcellular location of DAPK relative to MTs or PTRN-1 has not been examined. To further address whether DAPK-1 might affect PTRN-1 or MTs directly, we examined the localization of functional GFP::DAPK-1. Unexpectedly, GFP::DAPK-1 formed discrete puncta that underwent rapid, directed movement within the larval or adult epidermis (dpy-7 promoter, Figure 7A-D, Video 5; col-19 promoter, Videos 6, 7). In the lateral epidermis GFP::DAPK-1 puncta moved along the anteroposterior body axis, occasionally reversing direction ( Figure 7C). In contrast, puncta in the dorsoventral epidermis exhibited a bias in movement away from the lateral epidermis ( Figure 7B,D). GFP::DAPK-1 puncta also aggregated at the edges of the epidermal ridges along the body ( Figure 7A), as well as in the anterior epidermis, where morphological defects were prominent (Figure 7-figure supplement 1A).
MT-dependent transport within the mature C. elegans epidermis has not been previously characterized. To understand whether other cellular components are transported within the epidermis we examined an early endosome marker, GFP::RAB-5, as endocytosis has been implicated in innate immune responses (Dierking et al., 2011). GFP::RAB-5 formed motile puncta that were smaller and slower-moving than those of GFP::DAPK-1, and moved in both directions in the dorsoventral epidermis (Video 8; Figure 7F,G), suggesting GFP::DAPK-1 dynamics are distinct from endosomal transport.
We asked whether DAPK-1's transport correlated with its functions in epidermal morphology. DAPK-1, like mammalian DAPK, contains an N-terminal kinase domain, a Calcium/Calmodulin binding domain, a set of ankyrin repeats, a P-loop, a cytoskeleton-interacting domain and a C-terminal death domain (Shiloh et al., 2013) (Figure 1-figure supplement 1A). Constructs lacking any one of these domains, or containing a K57A point mutation predicted to abolish kinase activity (Deiss et al., 1995), were unable to rescue dapk-1(ju4) nor did they induce dapk-1-like phenotypes (Figure 7-figure supplement 1C, Table 4). Uniquely, constructs lacking the kinase domain (DKinase) strongly enhanced the Mor phenotype of a partial loss of function allele dapk-1(ju469) (Figure 7-figure supplement 1C), and caused lethality in a ju4 mutant background ( Table 4). The enhancement of dapk-1 phenotypes suggests DAPK-1(DKinase) inhibits DAPK-1 in a dominant-negative manner and may also inhibit parallel pathways.

Discussion
C. elegans DAPK-1 acts as a negative regulator of epidermal maintenance and wound responses, through previously unknown mechanisms (Tong et al., 2009). Here, we identify a role for DAPK-1 in regulating the stability or the dynamics of the MT cytoskeleton. The MT minus end factor PTRN-1 is closely linked to DAPK-1 function. Moreover, DAPK-1 itself is transported along MTs, indicating a complex relationship between DAPK-1 and MT dynamics in epidermal growth and repair.

Normal and mutant functions of DAPK-1
We screened for suppression of morphological defects of the dapk-1(ju4) allele, which results in a missense S179L alteration in the peptide-binding ledge of the DAPK-1 kinase domain. This allele, unlike dapk-1(gk219) or deletions of the dapk-1 locus, causes fully penetrant morphological phenotypes. Our identification of stop codon mutations as intragenic revertants, as well as the finding that transgenic expression of S179L DAPK-1 induces Mor phenotypes, suggests dapk-1(ju4) is a gain of function mutation.
How the S179L change causes a gain of function remains to be determined. One hypothesis for why dapk-1(ju4) has a more penetrant effect on epidermal morphology than the complete loss of dapk-1 function is that the S179L change alters the phosphorylation specificity of DAPK-1. Like mammalian DAPK family members, DAPK-1 likely has many phosphorylation targets, some of which may affect epidermal morphology. One clue to the effect of S179L on protein function comes from our structure function study. Truncated versions of mammalian DAPK1 can have dominant-negative behavior (Cohen et al., 1997(Cohen et al., , 1999Lin et al., 2007). In our assays, DAPK-1 (Dkinase) strongly enhanced, but did not induce, dapk-1 phenotypes, while DAPK-1 (kinase dead) did neither. Thus, DAPK-1(S179L) is unlikely to be kinase-dead, but may have an aberrantly regulated kinase activity. Inappropriate activity of DAPK-1 targets in dapk-1(ju4) mutants may account for the stronger morphological defects in this genetic background.
Upregulation of the epidermal innate immune response observed in dapk-1(ju4) is genetically separable from epidermal morphology defects (Tong et al., 2009) (Figure 2A,B). Our analysis using MT drugs shows that the concentrations of drugs sufficient to suppress or induce morphological defects are insufficient to suppress or induce innate immune responses, consistent with previous observations (Zhang et al., 2015). Thus, epidermal morphology is more sensitive to MT disruption than is the innate immune response, and MT stabilization may affect innate immune responses indirectly via effects on epidermal integrity.

PTRN-1's function in epidermal development and the MT cytoskeleton
The Patronin/CAMSAP family of MT minus end-binding proteins has become the focus of intense investigation in several organisms (Akhmanova and Hoogenraad, 2015). The sole C. elegans member of this family, PTRN-1, is dispensable for epidermal development in part due to a parallel pathway involving g-tubulin and the Ninein-related protein NOCA-1 . ptrn-1(0) single mutant larvae show grossly wild-type epidermal MT plus-end dynamics and have slightly fewer circumferential MT bundles in the epidermis , a phenotype that worsens in adults (this work). Thus during postembryonic development the epidermis becomes more dependent on PTRN-1 and on noncentrosomal MT arrays. Although ptrn-1 is largely dispensable for epidermal morphology, its requirement becomes evident in wound repair in adults, paralleling the role of  PTRN-1 in adult axon regeneration in C. elegans (Chuang et al., 2014). The requirement for PTRN-1 in repair but not in development may reflect the prominent role of noncentrosomal MT arrays in mature neurons and epithelia. The MT-binding PTRN-1 CKK domain is not only sufficient to rescue ptrn-1(0) suppression phenotypes, but can induce epidermal defects when overexpressed. Depending on the particular CAMSAP family member, the CKK domain can target MT minus ends, or can bind along the lattice (Goodwin and Vale, 2010;Jiang et al., 2014). In C. elegans PTRN-1 requires the CKK domain to localize to MTs in neurons and muscles (Richardson et al., 2014), and the CKK domain can stabilize neuronal MTs (Chuang et al., 2014), consistent with MT stability being a determinant of epidermal morphology.
Little is known about how CKK domain activity is regulated, but our data suggest DAPK-1 might directly or indirectly inhibit CKK activity; moreover, the CH or CC domains of PTRN-1 appear to inhibit CKK activity. As the CC domain of PTRN-1 appears to target to minus ends, a possible mechanism is that in the absence of the CC domain, ectopic association of the CKK domain along the MT lattice (long thin filaments) results in aberrant MT stabilization. DAPK-1 might inhibit the activity of the CKK domain indirectly via the CC or CH domains (Figure 8).

DAPK-1 and MT-dependent transport in the epidermis
To our knowledge, DAPK family members have not been previously shown to undergo MT-dependent transport. Many questions remain concerning the mechanism of DAPK-1 transport; here we speculate on the possible function of DAPK-1 motility in the C. elegans epidermis, and why aberrant DAPK-1 function should trigger altered epidermal morphology.
In the C. elegans epidermis the MT cytoskeleton is required for cell shape changes (Priess and Hirsh, 1986;Quintin et al., 2016) and for nuclear migrations (Starr, 2011). Epidermal MTs are critical for larval development: noncentrosomal arrays of the larval epidermis require PTRN-1 and the NOCA-1/g-tubulin pathway . In contrast, PTRN-1 is required non-redundantly for dapk-1 epidermal defects, suggesting that in this context PTRN-1 and NOCA-1 pathways are not equivalent.  A key question is why MT stabilization should cause progressive cuticle hypertrophy in specific regions of the epidermis. We speculate that the architecture of the epidermis may make it increasingly dependent on intracellular transport as it grows during larval development and adult life. Epidermal nuclei are confined to lateral and ventral ridges. The squamified dorsoventral epidermis is a barrier to diffusion. Thus, as in neuronal axons, extensive intracellular transport may be required to supply distant regions of the cell with cellular constituents. DAPK-1 may maintain MT dynamics required for active transport of materials to the remoter areas of the epidermis. Furthermore, our data show that the loss of dapk-1 function has a stronger effect on MT architecture in the lateral epidermis compared to the dorsoventral epidermis, perhaps due to differences in the local MTOCs or PTRN-1 in these different compartments. Cell type-specific differences in CAMSAP function have been recently been reported in Drosophila. The spectraplakin Shot and Drosophila CAMSAP Patronin co-localize, maintain MT arrays, and act to generate cell polarity; however, in the oocyte, Shot and PTRN-1 appear to act in the same pathway (Nashchekin et al., 2016), whereas in follicular epithelial cells they appear to act in parallel (Khanal et al., 2016). Possibly, PTRN-1 or non-centrosomal MTOCs in the lateral epidermis are more dependent on DAPK-1 function than in the dorsoventral epidermis.
In conclusion, we have revealed an unexpected regulatory interaction between DAPK-1 kinase and the MT cytoskeleton in epidermal development and wound responses. Many questions remain to be explored, especially whether PTRN-1 or other MT-associated proteins are direct substrates of DAPK-1. More broadly, the mechanisms and roles of intracellular transport within the epidermis could be a model for intracellular transport in other syncytial tissues.

Analysis of antimicrobial peptide reporter expression
For visual comparisons of Pnlp-29-GFP/Pcol-12-dsRed (frIs7) expression we took images on a Zeiss compound microscope using a GFP long-pass filter set. To quantify Pnlp-29-GFP (frIs7) expression levels we used the COPAS C. elegans BIOSORT (Union Biometrica, Holliston MA) at the Troemel lab (UCSD). We calculated the ratio of green fluorescence to the internal control Pcol-12-dsRed.
(2) We measured MT bundle length by drawing a line scan along a circumferential MT, from the edge of the lateral epidermis or the dorsal or ventral midline, to the end of the bundle. MT bundles detached from the epidermal ridges were not measured. (3) We counted MT bundles in a 40 mm ROI along the lateral epidermis along the anteroposterior axis. (4) We counted MT bundles crossing a 40 mm line in the dorsoventral region, including bundles not attached to the lateral or dorsoventral networks. For parameters 2-4 the ROI begins 40 mm behind the nose tip and extends 40 mm posteriorly. Schematics of ROIs are in Figure 4B. Quantitation of GFP::PTRN-1 filaments (Pdpy-7-GFP::PTRN-1, ltSi541): (1) Number of filaments in the dorsoventral regions of the head were counted in an ROI extending 40 mm from the nose tip ( Figure 6B).
(2) Filaments in the lateral head epidermis were counted in a 200 mm 2 ROI, beginning 40 mm behind the nose. (3) Filaments were counted in the anterior lateral epidermis in a 200 mm 2 ROI. A line of GFP > 1 mm was considered a filament. We did not examine the defective region to avoid quantifying changes in localization secondary to the aberrant epidermal morphology.

Statistics
All statistical analyses were performed with GraphPad Prism.