Conserved NIMA kinases regulate multiple steps of endocytic trafficking

Human NIMA-related kinases have primarily been studied for their roles in cell cycle progression (NEK1/2/6/7/9), checkpoint–DNA-damage control (NEK1/2/4/5/10/11), and ciliogenesis (NEK1/4/8). We previously showed that Caenorhabditis elegans NEKL-2 (NEK8/9 homolog) and NEKL-3 (NEK6/7 homolog) regulate apical clathrin-mediated endocytosis (CME) in the worm epidermis and are essential for molting. Here we show that NEKL-2 and NEKL-3 also have distinct roles in controlling endosome function and morphology. Specifically, loss of NEKL-2 led to enlarged early endosomes with long tubular extensions but showed minimal effects on other compartments. In contrast, NEKL-3 depletion caused pronounced defects in early, late, and recycling endosomes. Consistently, NEKL-2 was strongly localized to early endosomes, whereas NEKL-3 was localized to multiple endosomal compartments. Loss of NEKLs also led to variable defects in the recycling of two resident cargoes of the trans-Golgi network (TGN), MIG-14/Wntless and TGN-38/TGN38, which were missorted to lysosomes after NEKL depletion. In addition, defects were observed in the uptake of clathrin-dependent (SMA-6/Type I BMP receptor) and independent cargoes (DAF-4/Type II BMP receptor) from the basolateral surface of epidermal cells after NEKL-2 or NEKL-3 depletion. Complementary studies in human cell lines further showed that siRNA knockdown of the NEKL-3 orthologs NEK6 and NEK7 led to missorting of the mannose 6-phosphate receptor from endosomes. Moreover, in multiple human cell types, depletion of NEK6 or NEK7 disrupted both early and recycling endosomal compartments, including the presence of excess tubulation within recycling endosomes, a defect also observed after NEKL-3 depletion in worms. Thus, NIMA family kinases carry out multiple functions during endocytosis in both worms and humans, consistent with our previous observation that human NEKL-3 orthologs can rescue molting and trafficking defects in C. elegans nekl-3 mutants. Our findings suggest that trafficking defects could underlie some of the proposed roles for NEK kinases in human disease.

C. elegans has four NIMA-related kinases, NEKL-1, NEKL-2, NEKL-3, and NEKL-4. NEKL-2, a homolog of human NEK8 and NEK9, contains a highly conserved kinase domain and a short disordered region at the C terminus [34]. NEKL-2 is expressed in a punctate pattern in the large epidermal syncytium, hyp7, where it colocalizes with the conserved ankyrin repeat proteins MLT-2 (ANKS6 homolog) and MLT-4 (INVS homolog). MLT-2 and MLT-4 are required for the proper localization of NEKL-2, and their physical association is conserved in vertebrates [34][35][36]. NEKL-3, a close homolog of human NEK6 and NEK7, contains a kinase domain only and is also expressed in a punctate pattern within hyp7. NEKL-3 forms a complex with a third conserved ankyrin repeat protein, MLT-3/ANKS3, which is required for the proper localization of NEKL-3 in C. elegans and NEK7 in vertebrates [37]. We have shown that NEKL-2 and NEKL-3, along with their MLT binding partners, are required for the completion of molting in C. elegans, as a strong loss of function in any of these proteins leads to molting defects in early larval stages [34,[38][39][40]. Moreover, mutation of the kinase domain of nekl-2 or nekl-3 leads to uniform molting defects indicating that the catalytic activity of the NEKLs is essential for the promotion of molting [34,39].
Molting is an essential feature for C. elegans to develop through its four larval stages and to reach adulthood. During molting, the old cuticle, an apical extracellular matrix (aECM) made up of proteins and lipids, detaches from the underlying epidermis and is broken down as a new cuticle is synthesized underneath. This process of aECM remodeling requires the trafficking of macromolecules required for the synthesis, degradation, and recycling of cuticle components [41,42]. We previously reported that NEKL-2 and NEKL-3 regulate clathrin-mediated endocytosis (CME) at the apical membrane of hyp7. Our findings indicated that loss of NEKL-2 or NEKL-3 leads to a defect in the uncoating of apical clathrin-coated vesicles, thereby preventing the flow of apical cargo through the endocytic pathway. Consistent with this, an apically expressed low-density lipoprotein-like receptor, LRP-1/Megalin, was observed to be trapped at or near the apical membrane in NEKL-depleted animals [34,43]. LRP-1 internalization by hyp7 is essential for molting, as it is required for the uptake of cholesterol, a substrate required for hormonal signaling pathways that coordinate the molting cycle [41,44,45].
The involvement of NEKLs in CME is also strongly supported by our genetic screens, which identified subunits of the clathrin adapter protein complex, AP2, as strong suppressors of both molting and trafficking defects in nekl-2 and nekl-3 mutants. Likewise, loss of FCHO-1, an allosteric activator of AP2, alleviates molting and trafficking defects in nekl mutants [43]. Our collective data indicate that NEKL-2 and NEKL-3 directly or indirectly promote clathrin uncoating and suggest that AP2 activity can also affect the uncoating process. However, a possible role for NEKLs in other aspects of intracellular trafficking was not explored.
In this study, we further investigated the role of NEKLs at different membranes and in different compartments within the endocytic network. Importantly, we found that NEKL-2 and NEKL-3 have distinct functions within different endosome types and are required for basolateral trafficking and cargo recycling. Importantly, we extended our studies to human cell lines and demonstrated that the human NEKL-3 orthologs, NEK6 and NEK7, also appear to function at multiple points within the endocytic pathway in mammals. These studies provide new insights into the biological roles of NEK/NEKL family members, which may have relevance to their roles in human development and disease.

NEKLs are required for normal endosome morphologies
Given the observed colocalization of NEKLs with endosomes, we next wanted to determine if the loss of NEKLs led to an observable defect in the morphologies of endosomal compartments. During the process of endocytosis, endosomes undergo maturation, which is accompanied by changes in shape, molecular composition, and function. In the case of early/sorting and recycling endosomes, tubular extensions extend and pinch off from larger vesicles, allowing proteins to be recycled to specific destinations such as the plasma membrane and trans-Golgi [52-55]. Disruptions to endosomal functions can lead to a wide variety of defects including alterations in cargo localization and abundance.
To avoid indirect consequences caused by molting defects, we assessed the roles of NEKLs on endosomal morphology using the auxin-inducible degron (AID) system, in which endogenously tagged NEKL-2::AID and NEKL-3::AID were depleted in day-1 adults after the addition of auxin [43, 56,57]. Depletion of NEKL-2::AID led to an increase in the overall size of the early endosome compartment (based on total intensity and area measurements; P rab-5 ::GFP::RAB-5) and, most strikingly, to an increase in the extent of tubulation (Fig 2A, 2B and 2D-2F). This phenotype suggests a requirement for NEKL-2 in the process of vesicle fission at early endosomes. In contrast, loss of NEKL-2 had more modest effects on the late endosome compartment (P hyp7 ::GFP::RAB-7), consistent with the observed strong colocalization of NEKL-2 to early, but not late, endosomes ( Fig 2G, 2H and 2J-2L). Specifically, although the mean number of vesicles was decreased in NEKL-2-depleted worms as compared with wild type, the morphology of late endosomes was not strongly affected, and the observed changes could be due in part to gross alterations in the early endosome caused by NEKL-2 depletion.
In the case of NEKL-3::AID, depletion led to strong defects at both early and late endosomes, consistent with the localization of NEKL-3 to both compartments. In the case of early endosomes, however, the effects of NEKL-3 loss differed markedly from those after loss of NEKL-2 ( Fig 2B-2F). Rather than increased tubulation, loss of NEKL-3 led to increased variability in the size and shape of early endosomes accompanied by an increase in total signal intensity. Thus, while NEKL-2 and NEKL-3 both act at early endosomes, their roles in controlling early endosome morphology appear to be distinct. Depletion of NEKL-3 also led to a strong overall decrease in the abundance of late endosomes, which was associated with a decrease in total intensity, vesicle size, and changes in vesicle morphology (Fig 2G and 2I-2L). These defects may be due to a combination of the roles for NEKL-3 in directly regulating late endosomes and the potential secondary effects associated with its effects on early endosomes. We also note that even in the absence of auxin, NEKL-3::AID worms are smaller than wildtype worms and show some arrest due to molting defects, which may be attributable to a partial LOF caused by the AID tag [43].
We also examined the effects of NEKL loss on recycling endosomes, which were marked with endogenously (CRISPR) tagged RAB-11. Whereas NEKL-2::AID depletion had little or no effect on mScarlet::RAB-11 localization, we observed a strong increase in tubulation after loss of NEKL-3:AID, which was accompanied by an increase in total intensity and vesicle size (Fig 2M-2R). For technical reasons, we were unable to carry out colocalization studies between the NEKLs and either mScarlet::RAB-11 or GFP::RAB-11. Our observations, however, suggest that NEKL-3 may also function at recycling endosomes, possibly to control vesicle fission. In contrast to endosomal compartments, depletion of NEKL-2::AID and NEKL-3::AID had little or no effect on the morphology of the Golgi compartment, based on a P hyp7 ::AMAN-2:: mNeonGreen marker (S3 Fig). Nevertheless, the total number of AMAN-2-positive vesicles was variably reduced in these backgrounds, with strongest effects shown for NEKL-3::AID (S3A- S3C Fig), which could be caused by reduced endosome-to-Golgi recycling. Consistent with an indirect effect, a multi-copy NEKL-3::mCherry reporter [34] showed very low levels of colocalization with AMAN-2:: mNeonGreen (S3D Fig). Collectively, our studies demonstrate a role for NEKL-2 and NEKL-3 in regulating endosomal morphologies and further suggest direct functions within endosomal compartments. , and perimeter of the vesicles (R) were plotted for individual worms expressing mScarlet::RAB-11. Area is in square pixels; perimeter is the number of pixels in the boundary of the object. Error bars represent the 95% confidence intervals. p-Values were obtained by comparing means using an unpaired t-test: ****p < 0.0001, ***p < 0.001, *p < 0.05; ns, not significant (p > 0.05). Raw data are available in S1 File. https://doi.org/10.1371/journal.pgen.1010741.g002
To determine whether NEKL kinases are also required for trafficking at basolateral membranes of hyp7, we examined the internalization of two basolaterally localized BMP-family receptors. SMA-6 (Type I BMP receptor) and DAF-4 (Type II BMP receptor) form a heteromeric complex that binds to DBL-1 (BMP ligand), leading to the regulation of genes controlling C. elegans body size and male mating structures [65]. The internalization of SMA-6 depends on CME, as loss of AP2 function leads to the accumulation of SMA-6 at the basolateral membrane of intestinal cells [66]. In contrast, DAF-4 internalization does not require AP2, suggesting that it is endocytosed through a clathrin-independent mechanism [66].
To investigate possible roles for NEKLs in basolateral endocytosis, we expressed SMA-6:: GFP and DAF-4::GFP using a hyp7-specific promoter and examined expression at the membrane of hyp7 adjacent to the seam cell (referred to hereafter as the lateral membrane) and at the membrane that forms the interior boundary of hyp7 (referred to hereafter as the basal membrane; Fig 3A and 3B). Notably, the mean intensity of P hyp7 ::SMA-6::GFP was increased by~1.3-fold or~2.3-fold when NEKL-2 or NEKL-3 were depleted, respectively ( Fig 3C-3E and 3H). Moreover, a strong increase (~4.8-fold) in the total level of P hyp7 ::DAF-4::GFP was observed after NEKL-3::AID depletion (Fig 3F, 3G and 3I), suggesting that NEKL-3 may also be required for non-clathrin-dependent endocytosis or at the downstream junction of these pathways. In contrast, P hyp7 ::DAF-4::GFP was not detectably affected by depletion of NEKL-2 ( Fig 3H and 3J), suggesting that NEKL-2 may not be required for non-CME or may have a more limited role in endocytosis at the basolateral membrane relative to NEKL-3. Accumulation of these markers after NEKL-3 depletion was most evident at or near the basal hyp7 membrane, indicating a defect in plasma membrane uptake and/or processing through the early steps of endocytosis. In addition, some accumulation at or near the basolateral membrane with seam cells was also observed (S4 Fig). Together these results show that NEKL-2 and NEKL-3 play variable roles in the endocytosis of basolateral BMP receptors, although much stronger effects were observed for NEKL-3 in this process.

NEKLs are required for cargo sorting
Membrane cargoes that are internalized by endocytosis are first delivered to sorting/early endosomes after which they may be recycled back to their original compartment(s) or routed to the lysosome for degradation. Previous studies have characterized the recycling routes for two conserved cargo proteins, MIG-14/Wntless and TGN-38/TGN38, both of which originate from the trans-Golgi [67][68][69][70][71]. MIG-14 is a transmembrane protein that binds to Wnt ligands in the Golgi and is responsible for delivering Wnts to the plasma membrane for secretion. Once at the plasma membrane, MIG-14 is endocytosed and recycled back to the Golgi through a retrograde pathway that requires the retromer complex [72]. In the absence of retromer function, MIG-14 is missorted to the lysosome and is degraded [67,68]. Likewise, TGN-38 is a trans-Golgi resident that is also recycled from the plasma membrane to the trans-Golgi by a retromer-associated retrograde pathway [67,73,74].
To test if NEKL depletion leads to the aberrant degradation of cargoes, we used CRISPR methods to introduce a partial loss-of-function mutation into cup-5/trpml1, which encodes a conserved channel protein required for the maturation of lysosomes and whose loss leads to reduced lysosomal function [75][76][77]. Notably, reduction of CUP-5 activity led to an~1.5-fold increase in the total expression of P hyp7 ::TGN-38::GFP in both NEKL-2::AID and NEKL-3:: AID depleted strains (Fig 4I-4K). Moreover, accumulation of P hyp7 ::TGN-38::GFP was detected in moderate-to-large internal structures in these strains, which may correspond to defective lysosomes. Together these results indicate that, in addition to functioning in membrane cargo uptake, NEKL-2 and NEKL-3 are also required for the correct sorting and recycling of internalized cargo. We note that loss of cup-5 in an otherwise wild-type background also led to the accumulation of P hyp7 ::TGN-38::GFP, suggesting that some portion of TGN-38 is normally delivered to the lysosome in wild-type animals and that it's degradation depends on lysosomal function (S5 Fig). TGN-38 accumulations, however, appeared larger in cup-5 were omitted for clarity of presentation. Error bars represent the 95% confidence intervals. Statistical significance was determined using a two-tailed, unpaired t-test: **p < 0.01, ****p < 0.0001. Raw data are available in S1 File.
Depletion of NEKL-3::AID but not NEKL-2::AID also led to a reduction in the levels of P hyp7 ::MIG-14::GFP at the hyp7 membrane surrounding the ALM and PLM neurons, which run longitudinally through hyp7 along the anterior-posterior axis (S6 Fig). Altogether our results indicate that loss of NEKLs leads to a range of defects affecting endosomal compartments as well as cargo internalization and sorting. Moreover, although nekl-2 and nekl-3 Error bars represent the 95% confidence intervals. Statistical significance was determined using a two-tailed, unpaired t-test: ****p < 0.0001; ns, not significant (p > 0.05). Raw data are available in S1 File.
https://doi.org/10.1371/journal.pgen.1010741.g004 mutants appear superficially identical with respect to molting defects, they act in largely distinct subcellular compartments and lead to different subcellular phenotypes when depleted.

The mammalian homologs of NEKL-3, NEK6 and NEK7, are required for normal endocytic trafficking in human cells
Human NEK6 and NEK7 are both~70% identical and~85% similar to NEKL-3 and can rescue molting and trafficking-associated defects when expressed in C. elegans nekl-3 mutants [43]. Nevertheless, a systematic examination of potential roles for NEK6 and NEK7 in intracellular trafficking has not been undertaken. Using siRNA approaches, we were able to substantially reduce protein levels of NEK6 and NEK7 within 48 h of oligonucleotide transfection into cell lines (Fig 5A). As a first measure, we examined the effects of NEK6 and NEK7 knockdown on the cation-independent mannose 6-phosphate receptor (M6PR) in HeLa cells grown at steady state. M6PR is responsible for the delivery of mannose 6-phosphate-tagged lysosomal hydrolases from the trans-Golgi network (TGN), and thus cycles rapidly between the TGN and late endocytic organelles. Accordingly, M6PR is highly sensitive to disruptions in endocytic trafficking, which lead to alterations in its distribution.
Relative to control cells, the mean fluorescence intensity of M6PR-containing endosomes showed a significant decrease (~3-fold) in both NEK6 and NEK7 knockdown cells (Fig 5B-5G and 5K). These findings suggest that, analogous to observations for MIG-14 and TGN-38 in C. elegans, M6PR may be aberrantly targeted for lysosomal degradation after loss of the NEKL-3 orthologs. In addition, the distribution of M6PR in NEK6/7-depleted cells was altered, with M6PR showing a 2-to 3-fold increase in the area of dispersal (Fig 5H-5J and 5L). Although this increased M6PR dispersal might reflect slightly larger cell sizes in the NEK-depleted cells, M6PR appears to be more homogeneously distributed in the knockdown cells, an effect that has been associated with defects in endosome-to-Golgi transport via the retromer complex [78]. These data imply that both NEK6 and NEK7 are involved in the regulation of endosometo-Golgi retrograde trafficking.
Given that loss of NEKLs led to a major impact on endosomes, we next assessed the impact of NEK6 and NEK7 depletion on EEA1, a well-characterized early endosomal protein that binds to endosomal PI3P via its FYVE domain [79][80][81]. Upon siRNA-mediated depletion of NEK6 or NEK7 ( Fig 6A) and immunostaining with antibodies against EEA1, endosomal size was increased in cells depleted of NEK6 or NEK7, with those lacking NEK7 displaying especially large ring-like structures decorated by EEA1 (Fig 6B-6G and 6H). In contrast, the number of EEA1 endosomes was not significantly altered upon depletion of either NEK6 or NEK7 (Fig 6I). As a further test, we performed NEK6/7 siRNA knockdown in glioblastoma cells and observed an increase in the size of EEA1-marked vesicles after depletion of NEK7 but not NEK6 (S7 Fig), suggesting cell type-specific requirements for NEK functions. Together, these data suggest a role for NEK6 and NEK7 at endosomes, potentially in the regulation of fusion and/or fission events.
To further examine the role of NEK6 and NEK7 at endosomes, we analyzed the effect of their depletion on tubular recycling endosomes marked by the protein MICAL-L1 [82][83][84][85]. After depletion of NEK6 or NEK7 with siRNA oligonucleotides (Fig 7A), HeLa cells were fixed and immunostained with antibodies against endogenous MICAL-L1. Inhibition of NEK6 or NEK7 led to an increase in the MICAL-L1 tubular endosomal surface area relative to controls (Fig 7B-7E). Similar observations were observed for glioblastoma cells, with depletion of either NEK6 or NEK7 leading to increased tubulation (S8 Fig). Together, these data indicate roles for NEK6 and NEK7 in endosomal morphology and function in mammalian cells and support the possibility of their involvement in fusion and/or fission activities.

NIMA kinases are required for proper endosome morphology
In this study we have demonstrated that C. elegans NEKLs and human NEKs are important for maintaining the proper morphology of endosomal compartments. Loss of NEKL-2 led to enlarged early endosomal compartments with noticeably longer tubular extensions. These changes could indicate a failure of early endosomes to undergo fission, an expansion of early endosomes caused by abnormal cargo retention, or both. Notably, the effect of NEKL-3 loss on early endosomes was distinct from that of NEKL-2, as no increase in tubular extensions was observed. The shape of early endosomes, however, became irregular, and the mean intensity of the early endosomal marker GFP::RAB-5 was increased. This suggests that NEKL-2 and NEKL-3 have distinct roles in regulating early endosome morphology. Notably, siRNA-mediated knockdown of human NEK6 or NEK7 also led to an increase in the size of EEA-1-positive early endosomes in HeLa cells. However, whereas knockdown of NEK7 led to an increase in the size of early endosomes in glioblastoma cells, inhibition of NEK6 did not show this effect, suggesting that NEK6 and NEK7 may regulate early endosomal morphology in a cell type-specific manner.
Consistent with their distinct localization patterns, depletion of NEKL-3 had much stronger effects on late endosomal compartments than did loss of NEKL-2. We note, however, that effects on late endosomes could be due in part to perturbations of early endosomes caused by loss of NEKL-2 and NEKL-3. For technical reasons, we could not determine the expression of NEKLs in recycling endosomes. However, an increase in membrane tubulation was observed with recycling endosomes in hyp7 after loss of NEKL-3 but not NEKL-2. Collectively, these results indicate that NEKL-2 and NEKL-3 have important but distinct roles in regulating endosomal morphology.

NIMA kinases may broadly affect cargo uptake and recycling
We examined two Golgi residents, MIG-14 and TGN-38, to better understand how NEKLs may influence the endocytic recycling process. We observed a dramatic reduction in the MIG-14 and TGN-38 signal after depletion of NEKL-3. In contrast, loss of NEKL-2 led to a strong reduction in TGN-38, but only minimal effects on MIG-14. These findings suggest that specific cargoes are variably missorted to the lysosomal degradative pathway when NEKL-2 or NEKL-3 is depleted, as was previously reported for mutants affecting actin assembly and the sorting nexin, SNX3 [67,68]. Consistent with this, expression of TGN-38 was rescued when lysosomal function was inhibited in NEKL::AID strains. Interestingly, defects consistent with abnormal recycling were also observed for M6PR when NEK6 and NEK7 were depleted in HeLa cells, consistent with M6PR being incorrectly targeted to lysosomes. Together these findings suggest that NIMA family kinases may have conserved functions in the sorting of cargoes from early endosomes. Previously, we reported that NEKL-2 and NEKL-3 regulate endocytosis at the apical membrane of hyp7 [43]. In this study, we found that the internalization of a basolateral membrane cargo, SMA-6, was also affected in worms with NEKL-2 or NEKL-3 depletion. Our data also suggest that NEKL-3 may have a role in clathrin-independent endocytosis, as loss of NEKL-3 led to an accumulation of a putative clathrin-independent cargo, DAF-4, near the basolateral surface. One potential caveat to this interpretation is that in the presence of ligand, SMA-6 and DAF-4 would be predicted to form a complex [86], and thus SMA-6 or DAF-4 could potentially affect each other's uptake or retention. For example, membrane retention of endogenous SMA-6 could impact the localization and uptake of DAF-4::GFP. However, SMA-6 is expressed at very low levels [87,88], and thus it seems unlikely that endogenous plasma-membrane associated SMA-6 could indirectly lead to the strong membrane/juxta-membrane retention of the more highly expressed DAF-4::GFP. Altogether, these findings point to roles for NEKL-3 in both CME and non-CME.

Fig 5. Effects of NEK6 or NEK7 depletion on mannose 6-phosphate receptor trafficking in human cells. (A) siRNA knockdown of NEK6 and NEK7 was validated by western blotting of HeLa cells that were mock-transfected or transfected with oligonucleotides specific for NEK6 (left panel) or NEK7 (right panel). GAPDH was used as the loading control (lower panels). (B-G) Mock-transfected cells (B and enlarged area in C), NEK6 siRNA-transfected cells (D and enlarged area in E), or NEK7 siRNA-transfected cells (F and enlarged area in G) were plated on coverslips and immunostained with antibodies
We note that neither SMA-6 nor DAF-4 appeared to be incorrectly targeted to lysosomes after NEKL depletion, which differs from what we observed for MIG-14 and TGN-38. We speculate that SMA-6 and DAF-4 may be trapped at or near the basolateral surface in NEKL-3::AID depleted worms, as our previous study failed to detect clathrin accumulation at basolateral membranes of hyp7 [43]. Moreover, we failed to detect GFP::RAB-5 accumulation at or near basolateral surface in NEKL-3::AID depleted worms, suggesting that the basolateral accumulation of SMA-6 and DAF-4 is not within a RAB-5-marked early endosomal compartment. This buildup SMA-6 and DAF-4 could be due plasma membrane accumulation, internalized juxta-membrane vesicles, or a combination of the two.

What are the core functions of NEKLs in intracellular trafficking?
In a previous study, we reported that NEKL-2 and NEKL-3 regulate CME at the apical membrane and that they promote the uncoating of clathrin from internalized vesicles [43]. Our current study, however, suggests that the primary sites for trafficking regulation by NEKL-2 and NEKL-3 may be endosomes, and we failed to observe strong colocalization of NEKL-2 or NEKL-3 to AP2-decorated clathrin coated pits and vesicles. Thus, the clathrin-uncoating defects we previously observed could in part be an indirect consequence of endosomal defects. In this context, it is worth noting that RAB-5, which colocalized with both NEKL-2 and NEKL-3, partially localizes to clathrin-coated pits and promotes clathrin and AP2 uncoating [51,89]. Moreover, human RAB5 has been detected at both early endosomes and clathrincoated vesicles [46,90]. Thus, it remains possible that low levels of NEKL-2 and NEKL-3 may exert direct effects on clathrin uncoating, possibly in conjunction with RAB-5.
The effects we observed on multiple cargoes and compartments indicate that NEKL-2 and NEKL-3 may have widespread functions in intracellular trafficking. Likewise, effects on multiple compartments were observed after human NEK6 and NEK7 knockdowns. This could be due to NIMA kinases regulating distinct targets at different locations within the endocytic network. Alternatively, NIMAs may act on a smaller number of targets, which in turn have widespread functions in trafficking.

Mock-transfected cells (B and enlarged area in C), NEK6 siRNA-transfected cells (D and enlarged area in E)
, or NEK7 siRNA-transfected cells (F and enlarged area in G) were plated on coverslips and immunostained with antibodies against the early/sorting endosome marker protein EEA1. Yellow boxes in B, D, and F indicate the area of higher magnification shown in C, E, and G, respectively. (H) The mean size of EEA1-containing endosomes in mocktransfected, NEK6 siRNA-transfected, and NEK7 siRNA-transfected cells. (I) The number of EEA1-containing endosomes in mock-transfected, NEK6 siRNA-transfected, and NEK7 siRNA-transfected cells. Scale bar in B = 10 μm for B, D, and F; Scale bar in C = 5 μm for C, E, and G. Statistical significance was determined using Student's unpaired t-test. Raw data are available in S1 File. https://doi.org/10.1371/journal.pgen.1010741.g006

PLOS GENETICS
In this latter category, one compelling model is that NIMAs could regulate endosomal actin. Actin polymerization is deployed throughout endocytosis to provide mechanical forces needed to bend membranes, facilitate vesicle fission, and promote short-range vesicle transport [91][92][93][94][95]. For example, actin filaments are enriched at sorting endosomes and promote the fission of tubular extensions [52,96]. Actin assembly also plays a role in the sorting of macromolecules from early endosomes and their transport to the trans-Golgi [97,98]. Consistent with this idea, our previous studies showed that actin organization was strongly disrupted in nekl mutants and that nekl molting defects can be suppressed by loss of function in the conserved actin regulator CDC-42, along with one of its effectors, SID-3/ACK1/2 (activated CDC42 kinase) [38]. Moreover, activated (GTP-bound) CDC-42 was strongly upregulated in nekl mutants [38]. Interestingly, based on protein-interaction studies, CDC42 was reported to interact with both NEK6 and NEK7 and could therefore connect the NEKL-MLT network to actin polymerization and to their roles throughout the endocytic network. Future studies will seek to characterize the functional link between NIMA kinases and the actin cytoskeleton, along with their specific roles in regulating endocytosis.

Human NEKs in endocytosis and cancer
Although human homologs of NEKL-3 (NEK6, NEK7) and NEKL-2 (NEK9) have largely been studied for their roles in cell division, several studies have hinted at potential functions in endocytosis [99][100][101][102][103][104]. A high-throughput siRNA-based screen of mammalian protein kinases indicated that siRNA-mediated knockdown of NEK6 and NEK7 strongly disrupts CME in HeLa cells [99]. Moreover, genome-wide siRNA studies showed that knockdown of multiple NEK kinases, including NEK6, resulted in the abnormal uptake of endocytic cargoes [101]. A more recent study identified NEK6 as a potential downstream target of NRP-1, which is required for transferrin endocytosis by Trypanosoma brucei [105]. Moreover, proteomic studies of NEK6 and NEK7 identified interactions with several trafficking components including alpha and beta subunits of the AP2 adaptor complex (NEK7) [102,103]. Such reports are consistent with our current findings that NEK6 and NEK7 play roles in endocytic trafficking in human cells. Lastly, our ability to rescue both molting and trafficking defects through the expression of human NEKs in C. elegans nekl mutants suggests that these functions are conserved [43].
Finally, our studies provide evidence that NEKLs may affect two highly conserved signaling pathways, BMP and Wnt, which are misregulated in many human cancers [106][107][108]. Previous studies have described the role of endocytosis in cancers including how endocytosis adapts to the needs of cancer cells, a phenomenon known as "adaptive CME" [109][110][111]. For example, upregulation of clathrin light-chain (CLCb) leads to an increased rate of CME; this in turn alters the trafficking of oncogenic epidermal growth factor receptor (EGFR) and promotes cancer cell migration and metastasis [112]. Thus, it is possible that NIMA kinases, which are overexpressed in many tumors, could be driving cancer formation in part through their roles in trafficking as well as by their previously described functions in mitosis [19, [113][114][115][116][117]. Future studies will elucidate how NIMA kinases control trafficking, which will aid in understanding their involvement in human disease.

Strains and maintenance
C. elegans strains were maintained according to standard protocols [118] and were propagated at 22˚C unless stated otherwise. Strains used in this study are listed in S1 Table. Loss-of-function alleles of cup-5 (fd395 and fd397) were generated by the CRISPR-Cas9 protocol described by Ghanta et al. [119]. Sequences for the sgRNA and repair template are given below. Capital letters represent the altered or inserted nucleotides. sgRNA: 5 0 -aacgatgcgcttattatcat-3 0 Repair Template: 5 0 -aacttaaatttttataaaaattacctgaaactcgatggatatttttgcaacGGTCTCaatgataataagcgcatcgttgaccacaatcatcacataccacaaattcag-3 0

Image acquisition
Confocal fluorescence images in Figs 1-4 and S1 and S3 and S5 were acquired using an Olympus IX83 inverted microscope with a Yokogawa spinning-disc confocal head (CSU-W1). z-Stack images were acquired using a 100×, 1.35 N.A. silicone oil objective. cellSense 3.1 software (Olympus Corporation) was used for image acquisition. Fluorescence images in S2 Fig were acquired using an Olympus IX81 inverted microscope with a Yokogawa spinning-disc confocal head (CSU-X1). MetaMorph 7.7 software was used for image acquisition. z-Stack images were acquired using a 100×, 1.40 N.A. oil objective.
Confocal images for Figs 5-7 and S6 and S7 were acquired using an Olympus IX83 inverted microscope with a Yokogawa spinning-disc confocal head (CSU-W1) or with a Zeiss LSM 800 microscope with a 63×, 1.40 N.A. objective with appropriate filters.

Image analysis
Mean intensity (measured in arbitrary units, a.u.), quantification of various shape parameters, and the colocalization analysis were performed using Fiji software [122]. To quantify mean intensity (Figs 2D, 2J, 2P, 3H, 3I, 4D, 4H, 4K and S3B) for a z-plane of interest, rolling ball background subtraction was performed (radius = 50 pixels), and the polygon selection tool was used to choose the region of hyp7 in which the mean intensity was quantified.
To quantify shape parameters, a z-plane of interest was selected, and the minimum filter (radius = 10.0 pixels) was applied to the raw image. The filter-applied image was then subtracted from the raw image using the image calculator function. Images were then thresholded, and the Despeckle function was used to remove noise corresponding to 1 pixel in size. Finally, various shape parameters were quantified using the Shape Descriptor plugin.
For colocalization studies, the raw z-stack images were deconvoluted using the Wiener deconvolution algorithm (cellSense 3.1 software). The desired z-plane was extracted from both deconvoluted and raw z-stack images. Then, the Gaussian filter (radius = 10 pixels) was applied to the deconvoluted image, which was subtracted from the original deconvoluted image using the image calculator function. Next, these subtracted images were thresholded to obtain binary images to be used as masks. These binary masks were combined using the "AND" Boolean operation to the background-subtracted (rolling ball algorithm; radius = 50 pixels) raw images. Using the polygon tool, the region of hyp7 was selected in the combined images, and coloc2 plugin was used to calculate Manders' coefficient.
The mean area and intensity of fluorescence in HeLa cells Fig 5K and 5L was obtained with Zeiss LSM Zen software, after outlining a region of interest. EEA1-positive endosome size ( Fig  6H) was measured using Imaris software, and MICAL-L1 tubular endosome surface area ( Fig  7E) was quantified using ImageJ software., or the number of MICAL-L1 tubular recycling endosomes were counted manually (S8 Fig), since number and area are generally correlated. Endosome size in the human cell studies was quantified using NIH ImageJ. Size parameters were set for 0 -infinity. Brightness parameters were selected to eliminate recognition of background by ImageJ's particle counter while optimizing selection of true positive fluorescent pixels. Statistical analyses was performed in Prism by first testing the assumption of normal distribution with the D'Agostino and Pearson normality test. Statistical significance was then calculated with an unpaired two-tailed t-test.

Auxin treatment
Auxin (indole-3-acetic acid) was purchased from Alfa Aesar. A 100× stock auxin solution (0.4 M) was made by dissolving 0.7 g of auxin in 10 ml of 100% ethanol. A mixture of 25 μl of stock auxin solution and 225 μl of autoclaved deionized water was added to each plate containing day-1 adult worms.

Immunoblotting
HeLa cells were lysed in lysis buffer containing 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% NP-40, and 0.5% deoxycholate. Proteins from the lysates were separated by SDS-PAGE, transferred to nitrocellulose filters, and immunoblotted with antibodies using standard methods.

siRNA knockdown
Cells were transfected with siRNA oligonucleotides from Sigma (600 nM NEK6 600 or 100 nM NEK7) using Dharmafect transfection reagent (Dharmacon) for 48 h, prior to validation of knockdown by immunoblotting using GAPDH as the loading control as described above.