Centrioles degenerate early in neuronal differentiation

Next, we sought to examine at what stage of neuronal differentiation centrioles are lost. Reconstructing the timeline of centriole degeneration and concomitant cilia assembly is aided by the highly stereotypical nature of C. elegans development: the timing and pattern of cell divisions and the relative position of cells is essentially unchanged from animal to animal (Sulston et al., 1983), as is the timing of individual cell-cycle events (Oegema and Hyman, 2006; Begasse et al., 2015). This invariance greatly facilitates examination of multiple markers alongside ultrastructural analysis at different stages of differentiation. For our analysis, we decided to focus on the amphids, the primary sensory organ in the head of the animal, easily identified and examined by both light and electron microscopy.

The neuroblasts that give rise to anterior neurons including the amphids complete their terminal cell divisions after the end of gastrulation, between 330 and 450 min after fertilization (times at 22°C). Anterior neurons move toward the tip of the head and then posterior again, laying down their dendrites as they go, a process termed retrograde extension (Fig. 2 A and Video 3; Heiman and Shaham, 2009). The embryo elongates and becomes thinner because of contraction of the hypodermis from the so-called comma stage (430 min after fertilization) through the 1.5-fold (460 min), twofold (490 min), and threefold (550 min) stages (Fig. 2 A). The worm eventually hatches at 840 min after fertilization (Sulston et al., 1983).

Open in a separate window

Figure 2.

Centrioles are lost early in neuronal differentiation. (A) Stills from time-lapse video of embryo expressing myristoylated GFP (myrGFP) in amphid neurons undergoing retrograde migration. DIC and single-plane GFP images taken at the indicated time intervals corresponding to comma, 1.5-fold, twofold, and threefold stages. Note the position of dendritic tips (arrowheads) remains fixed as cell bodies (asterisks) move. At the threefold stage, the worm begins to move inside the egg because of muscle contraction. See also Video 3. (B) Tomographic slice and 3D reconstruction model of centriole in a myristoylated GFP–positive neuron of a comma-stage embryo. Note the centriole displays doublet microtubules. Electron densities at the center of the centriole may represent elements of the central tube/cartwheel (arrowhead). See also Video 4. (C and D) Immunofluorescence micrographs of comma-, 1.5-fold–, twofold-, and threefold-stage embryos expressing myristoylated GFP in amphid neurons and stained for HYLS-1 and SAS-6 (C) or SAS-4 (D). Insets show magnified view of ciliary base (1) and nonneuronal cells elsewhere in the head of the embryo (2). Centriolar signal is lost from the ciliary base, with loss of SAS-6 preceding that of SAS-4, while HYLS-1 remains. All three proteins are lost in nonneuronal cells coinciding with terminal differentiation. (E and F) Quantitation of centriolar signal in amphid neurons (percentage of HYLS-1 foci positive for SAS-6 or SAS-4; E) and nonneuronal cells (HYLS-1 foci per nucleus; F) from images as in C and D. Error bars are 95% confidence intervals. n = 5–9 animals per condition. Bars: (A, C, and D) 10 µm; (B) 50 nm; (C and D, insets)1 µm.

Shortly after terminal cell division at the comma stage, centrioles are still clearly present, as assessed by SAS-4 and SAS-6 signal coincident with HYLS-1 in neurons marked by Pdyf-7 neuronal promoter-driven GFP (Fig. 2, C and D). Interestingly, these centrioles appear to be composed of doublet microtubules (Fig. 2 B and Video 4), in contrast to the singlet microtubule centrioles found in sperm and the early embryo (Wolf et al., 1978; Pelletier et al., 2006). Over the course of the next hour from the comma to twofold stage, SAS-4 and SAS-6 signal declines dramatically, with loss of the cartwheel component SAS-6 preceding that of the centriolar microtubule wall component SAS-4 (Fig. 2, C–E). Dynamically localized centriolar components ZYG-1, SPD-2, and SAS-5 are likewise lost with a timing similar to or preceding SAS-6 (Fig. S2). In contrast, HYLS-1 signal remains largely unchanged and is used as a reference point for this analysis. Centriole degeneration is not unique to neuronal differentiation. Indeed, there is a marked loss of centriolar protein signal (in nonneuronal cells also including HYLS-1) across the entire embryo at the same time (Fig. 2, C–F), which coincides with the terminal divisions of most of the cells giving rise to the C. elegans soma (Sulston et al., 1983). Centriole degeneration then occurs relatively early during neuronal differentiation, raising the question how far cells have progressed in ciliogenesis at this stage.

Formation of ciliary structures

Transition zone assembly

The transition zone is one of the most striking and characteristic features of the cilium. At 1 µm in length compared with ~300 nm in vertebrates, the transition zone is rather well developed in C. elegans and appears to contain periodic elements which may correspond to stacked Y-links (Lambacher et al., 2016). While the molecular architecture of the transition zone remains to be fully elucidated, increasing evidence points to it being organized into multiple protein complexes or modules, recruited independently of each other and with distinct localization patterns and functions (Garcia-Gonzalo et al., 2011; Sang et al., 2011; Williams et al., 2011; Chih et al., 2012; Basiri et al., 2014; Schouteden et al., 2015; Pratt et al., 2016; Vieillard et al., 2016). Of these, the proteins of the MKS and NPHP modules localize to the transition zone periphery where they cooperate in the formation of Y-links (Williams et al., 2011; Yang et al., 2015; Lambacher et al., 2016). The contribution of CEP290 and RPGRIP1L is less clear, but their location at the proximal and distal core of the transition zone, respectively (Yang et al., 2015), supports a role in scaffolding and anchoring transition zone structures (Basiri et al., 2014; Schouteden et al., 2015; Li et al., 2016).

To examine the timing of transition zone assembly, we chose MKSR-2 and NPHP-4 as upstream components of the MKS and NPHP module assembly pathways involved in Y-link formation (Williams et al., 2011) and CCEP-290 as a putative central cylinder component (Fig. 3, A and B; Schouteden et al., 2015). Endogenous promoter GFP fusions were validated by their ability to rescue the respective null mutant in the dye-fill assay (Fig. S1 C). Recruitment of these proteins to the site of cilia assembly was then assessed using mCherry:HYLS-1 as a marker. For all three proteins, initial recruitment occurs at the twofold stage (Fig. 3 C), a stage at which SAS-4 but not SAS-6 is still present (compare Fig. 2, C–E). Levels subsequently increase, with the transition zone expanding to near-adult dimensions by the threefold stage (Fig. 3 C). Examination of twofold stage embryos by electron tomography revealed centrioles docked at the cell surface, decorated at their distal end with a single array of Y-links and an incomplete central cylinder (Fig. 3 D and Video 5). Consistent with the rather minimal signal for CCEP-290, MKSR-2, and NPHP-4 at this time, the transition zone does not yet extend significantly in length, with the incipient cilium capped within 100 nm of the centriole by the ciliary membrane. While the centriole itself still appears reasonably intact, there is already a significant widening at the proximal end, potentially reflecting the loss of SAS-6–containing inner centriolar structures. Also notable is an apparent daughter centriole, oriented orthogonally, not found in mature cilia.

Open in a separate window

Figure 3.

Transition zone assembly. (A and B) 3D reconstruction model (A) and molecular assembly hierarchies (B) for the C. elegans transition zone based on Schouteden et al. (2015) (A) and Huang et al. (2011), Williams et al. (2011), Jensen et al. (2015), Roberson et al. (2015), Schouteden et al. (2015), and Li et al. (2016) (B). CCEP-290 is an essential component of the central cylinder, whereas MKS and NPHP module components function in assembly of Y-links. MKSR-2 and NPHP-4 are upstream components in the MKS and NPHP assembly pathways, respectively. (C) Recruitment of transition zone components to the site of cilia assembly marked by HYLS-1. Panels show 1.5-fold–, twofold-, and threefold-stage embryos and L1-stage larvae coexpressing GFP:CCEP-290, MKSR-2, NPHP-4, and mCherry:HYLS-1. Initial recruitment of all three components occurs at the twofold stage, with signal reaching maximal levels at the threefold stage. (D) Tomographic slices and 3D reconstruction model of the ciliary base of twofold-stage amphid cilium. A pair of centrioles can be seen, one of which is decorated by a single array of Y-links and a discontinuous central cylinder and enveloped by the ciliary membrane. Note marked widening of the centriole toward the base. See also Video 5. Bars: (C) 10 µm; (C, insets) 1 µm; (D) 100 nm.

In summary, transition zone proteins begin to be recruited and key elements (Y-links and central cylinder) start to assemble at a time when centrioles are still present. However, transition zone expansion and axoneme extension occur at later stages, after loss of centrioles and centriolar proteins.

IFT-dependent axoneme extension

Axoneme extension beyond the transition zone involves microtubule motor-driven transport of IFT trains assembled from IFT-A and IFT-B subcomplexes delivering subunits to the ciliary tip (Cole et al., 1998). In C. elegans, anterograde transport is mediated by two kinesin-2 motors, heterotrimeric kinesin-II and homodimeric OSM-3. Both motors cooperate to form the proximal segment of the axoneme, composed of nine doublet microtubules, whereas assembly of the distal segment, composed of nine singlets, depends on OSM-3 alone (Snow et al., 2004). IFT docking at the ciliary base has been shown to require DYF-19/Fbf1, a protein associated with transition fibers in vertebrates (Wei et al., 2013).

To examine IFT-dependent axoneme extension during C. elegans development, we chose CHE-11 and OSM-6 as representatives of the IFT subcomplexes A and B, respectively, and KAP-1 (kinesin-II) and OSM-3 for the two anterograde motors (Fig. 4 A). For each protein, endogenous promoter fluorescent fusions were validated by mutant rescue where possible (kap-1 mutants do not present an easily detectable phenotype; Fig. S1 D). We then monitored recruitment using the transition zone marked by MKSR-2 and MKS-6 as a point of reference (Fig. 4 B). While some IFT proteins (KAP-1 and OSM-6) were observed as early as the twofold stage, significant accumulations of all four proteins were only observed after hatching at the first larval stage (L1). Coincident with initial IFT recruitment, DYF-19 was found to localize to the ciliary base from the twofold stage (Fig. S3 A). Consistent with the lack of axoneme extension beyond the transition zone observed by electron microscopy in twofold-stage embryos, any early IFT signal largely overlaps with the transition zone. Only from the L1 stage onwards are IFT proteins clearly seen distal to the transition zone.

Open in a separate window

Figure 4.

IFT-dependent axoneme extension. (A) Schematic of anterograde IFT machinery in C. elegans. Assembly of the proximal segment depends on the coordinated transport of IFT trains by kinesin-II and OSM-3 together, while assembly of the distal segment depends on OSM-3 alone. (B) Recruitment of fluorescently tagged IFT components (green in merge) to the site of cilia assembly marked by the transition zone components MKS-6 and MKSR-2 (red). Panels show 1.5-fold–, twofold-, and threefold-stage embryos and L1- and L3-stage larvae coexpressing CHE-11:mCherry, KAP-1:GFP, OSM-3:mCherry, or OSM-6:GFP and GFP:MKSR2 or MKS-6:mCherry as indicated. Initial recruitment of KAP-1 and OSM-6 occurs at the twofold stage of CHE-11 and OSM-3 at the threefold stage. IFT trafficking is only observed at L1. (C) Tomographic slice through base of the amphid channel in L1 larva, with magnified view and 3D reconstruction model of a single cilium at the level of the transition zone and proximal axoneme. Note cilium appears to terminate close to the transition zone. See also Video 6. Bars: (B) 10 µm; (B, insets) 1 µm; (C) 200 nm; (C, insets) 100 nm.

In mature C. elegans amphids, 10 cilia extend into a channel formed by the sheath and socket glia cells through which they are in contact with the external medium. With their transition zones embedded in the matrix at the base of the sheath channel, the proximal segment of the axoneme extends 4 µm from the transition zone, whereas the singlet microtubules of the distal segment continue for another 2.5 µm (Fig. 4 A; Perkins et al., 1986; Doroquez et al., 2014). Electron tomography of the base of the amphid channel in L1 larvae show the expected number of cilia. However, while some cilia are already fully developed, others terminate close to the transition zone, with doublet microtubules extending to different lengths. These cilia further lack a clear distal segment of singlet microtubules (Fig. 4 C and Video 6). Axoneme extension then is a gradual process, largely occurring during the larval stages.

Immunofluorescence and fixed imaging

Immunofluorescence experiments were performed as previously described (Oegema et al., 2001) using Cy3/Cy5 directly labeled affinity-purified rabbit polyclonal antibodies against HYLS-1 (Dammermann et al., 2009), SAS-4, SAS-5, SAS-6, SPD-2, and ZYG-1 (Dammermann et al., 2004), as well as unlabeled mouse monoclonal antibodies against polyglutamylated tubulin (GT335; provided by C. Janke, Institut Curie, Paris, France). Residual fluorescence signal was used to detect myristoylated GFP protein in fixed embryos. In brief, embryos were permeabilized by freeze crack, fixed in −20°C methanol for 20 min, rehydrated in PBS, blocked for 20 min in AbDil (PBS, 2% BSA, 0.1% Triton X-100), incubated with directly labeled antibodies at 1 µg/ml (4 µg/ml GT335) in AbDil for 1.5 h, washed with PBST (PBS, 0.1% Triton X-100), incubated with secondary antibody (if applicable, 15 µg/ml Cy5-labeled donkey anti–mouse antibody; Jackson ImmunoResearch Laboratories, Inc.) in AbDil for 1 h, incubated with 1 µg/ml Hoechst 33342 in PBS for 5 min, washed again in PBST, and mounted in 0.5% p-phenylenediamine, 20 mM Tris, pH 8.8, and 90% glycerol. 0.2-µm 3D wide-field datasets were acquired using an Olympus 100× 1.4 NA Super Plan Apochromat lens on a DeltaVision microscope (Applied Precision) equipped with a 7-Color SSI module and CoolSNAP-HQ2 cooled charged-coupled device (CCD) camera (Photometrics), computationally deconvolved using the enhanced ratio constrained iterative deconvolution algorithm, and maximum intensity projected in SoftWorx (Applied Precision), before being imported into Adobe Photoshop for panel preparation. Exposure conditions and scaling were kept constant when comparing signal across different developmental stages. No nonlinear gamma corrections were applied during image processing.

Live imaging

For live imaging, worms were anaesthetized with 10 mM tetramisole and mounted on 2% agarose pads. Embryos were similarly mounted on agarose pads. Imaging was performed at room temperature (20–22°C) using an Olympus 60× 1.42 NA PLAPON Apochromat objective and 1.6× optovar on the set-up described above. For examination of neuronal morphology and recruitment/loss of centriolar and ciliary markers, 3D wide-field datasets were acquired, deconvolved, and processed as described above. Exposure conditions and scaling were kept constant when comparing signal across different developmental stages. For time-lapse imaging of neurite extension, 0.6-µm z stacks of GFP and single-plane differential interference contrast (DIC) images of embryos expressing myristoylated GFP under a panneuronal promoter were acquired every 5 min, using low-power LED illumination to avoid photodamage. GFP z stacks were maximum intensity projected in SoftWorx (Applied Precision), and image sequences were imported into ImageJ (National Institutes of Health) for video preparation.

Electron microscopy and tomography

L4-stage worms were prepared by chemical fixation as previously described (Serwas and Dammermann, 2015). In brief, worms were fixed in 2.5% glutaraldehyde in cytoskeleton buffer (100 mM methyl ester sulfonate, 150 mM NaCl, 5 mM EGTA, 5 mM MgCl₂, and 5 mM glucose in ddH₂O, pH 6.1) overnight at 4°C. Samples were washed three times in the same buffer, postfixed for 30 min in 0.5% osmium tetroxide in buffer, and washed three times in buffer and once in ddH₂O. Finally, samples were dehydrated for 15 min each in 40, 60, and 80%, twice in 95%, and three times in 100% acetone. L1-stage worms were fixed by high-pressure freezing (HPF Compact 01; M. Wohlwend GmbH) and freeze substitution (EM AFS2; Leica Biosystems) using 5% BSA in M9 buffer as filler and an acetone-based freeze substitution medium containing 0.5% glutaraldehyde, 0.5% osmium tetroxide, and 0.2% uranyl acetate. Freeze substitution was started at −140°C, and then temperature was gradually increased to −90°C over 3 h, kept at −90°C for 61 h, gradually increased to −30°C over 20 h, kept at −30°C for 12 h, and finally increased to 0°C over 15 h. Embryos were prepared according to the procedure described by Kolotuev et al. (2010). In brief, adult worms expressing myristoylated GFP under a panneuronal promoter were dissected in 2 µl M9 buffer on an Aclar film (no. 50425; Electron Microscopy Sciences) placed on a glass microscope slide. 30 µl of 4% low–melting point agarose (no. 0815; Amresco) was added, and the sample sandwiched between a second Aclar film and glass slide using an additional Aclar film piece containing a rectangular hole (~3 × 1.5 cm) as a spacer. Embryos were excised using a biopsy punch and imaged by light microscopy to determine their developmental stage based on overall morphology by DIC and neuronal morphology by GFP fluorescence. Embryos of the desired stage were selected and fixed using high-pressure freezing (EM PACT; Leica Biosystems) and freeze substitution in 0.5% glutaraldehyde, 1% osmium tetroxide, and 0.2% uranyl acetate in acetone using the same time scheme as above. All samples were embedded in Agar100 resin after fixation and dehydration. For the data presented in Fig. 1 B, 70-nm serial sections were poststained with aqueous uranyl acetate and lead citrate and examined with a Morgagni 268D microscope (FEI) equipped with an 11-megapixel Morada CCD camera (Olympus) and operated at 80 kV. For all other experiments, 200-nm serial sections were poststained as above and examined with a Tecnai G2 20 microscope (FEI) equipped with an Eagle 4k HS CCD camera (FEI) and operated at 200 kV. Tilt series were acquired in single- or double-tilt axis geometry by stepwise tilting the sample from −60° to 60° in 1° increments using SerialEM (Mastronarde, 2005). Tomogram reconstruction with 10 nm gold beads as fiducials and model generation was performed using IMOD (Kremer et al., 1996). Tomographic slices presented a range from 10 to 25 nm in thickness.

Statistical analysis

No statistical method was used to predetermine sample size. For fluorescence microscopy data, reproducibility was confirmed in at least two independent experiments. For electron microscopy data, the number of animals examined is provided in Table S2. Statistical significant differences between samples were determined by applying Student’s t test in Prism (GraphPad Software). Data distribution was assumed to be normal, but this was not formally tested. Two-tailed p-values <0.05 were considered significant. All results are expressed as the mean ± 95% confidence intervals. Sample size (n) and p-values are indicated in the figure legends.

We thank Chris Campbell, Luisa Cochella, and Chad Pearson for comments on the manuscript; Daniela Nicastro, Piali Sengupta, the Vienna Biocenter C. elegans community, and members of the Dammermann laboratory for discussions; the Caenorhabditis Genetics Center, Dhanya Cheerambathur, Jinghua Hu, Carsten Janke, Michel Leroux, Shohei Mitani, Karen Oegema, Erwin Peterman, Amy Shaub Maddox, and Bradley Yoder for strains and reagents; and the Vienna Biocenter Core Facilities Electron Microscopy facility, especially Sonja Jacob and Harald Kotisch, for technical assistance.

This work was supported by start-up funding from Max F. Perutz Laboratories and the Austrian Science Fund (grant Y597-B20 to A. Dammermann), as well as a uni:docs predoctoral fellowship from the University of Vienna to T.Y. Su.

The authors declare no competing financial interests.

Author contributions: D. Serwas: conception and design, acquisition of data, analysis and interpretation of data, and drafting or revising the article; T.Y. Su, M. Roessler, and S. Wang: contributed unpublished essential data or reagents; and A. Dammermann: conception and design, analysis and interpretation of data, and drafting or revising the article.