Abstract
The primary cilium is an important regulator of signaling pathways in cell proliferation and differentiation. Mechanisms involved in cilium assembly and homeostasis are well described, but little is known about mammalian ciliary disassembly, which is required for cell cycle progression. We examined whether cilia disassembly in cultured mouse cells occurred by resorption into the cell or extracellular shedding. Live cell imaging of individual disassembly events revealed dynamic and heterogeneous behaviors, with rates varying by several orders of magnitude. Surprisingly, seconds-fast disassembly was the predominant method of cilium loss (83% of events), and we demonstrate that this is due to ciliary shedding. We tested the roles of candidate regulators of ciliary shedding, katanin and intracellular calcium. Katanin overexpression increased seconds-fast disassembly, and katanin and intracellular calcium levels independently, but not synergistically, reduced cilia length. This work provides new, detailed insights into mechanisms of primary ciliary disassembly in mammalian cells.
Introduction
Eukaryotic cilia and flagella are highly conserved organelles adapted to facilitate the interaction of cells with the surrounding environment. Many types of cilia are found in a wide diversity of organisms, tissues and cell types, with variable structural features underlying specific functions. Nevertheless, all cilia and flagella share a common structure consisting of a mature centriole (a basal body in this context), which nucleates a core of stable microtubule doublets (the axoneme), encased in a ciliary membrane distinct in composition from the plasma membrane. Contact between the basal body/axoneme and plasma membrane occurs near the base of the cilium at the transition zone, which is thought to segregate the ciliary compartment from the cytoplasm, thereby maintaining a specific population of proteins in the cilium (1–3).
The primary cilium occurs in single copy per cell in most vertebrate tissues, where it receives and transduces mechanical and molecular signals. In this role, primary cilia are essential for coordinating proliferative, metabolic, and developmental signaling pathways including Hedgehog, Receptor Tyrosine Kinase (RTK) growth factors (PDGF, TGF–β, IGF), GPCRs, and non-canonical Wnt(4–6). Furthermore, the primary cilium itself may contribute to cell cycle regulation (7–10). The importance of normal ciliary function is underscored by the diverse phenotypes associated with ciliopathies, developmental syndromes that occur when ciliary structure, function, or regulation are defective (11–13).
Primary cilia are closely intertwined with the cell cycle. In most proliferating animal cells, the primary cilium forms in G0/G1 and is disassembled prior to M-phase, releasing the centrioles to associate with the mitotic spindle (7,14). Cilium disassembly is linked to cell cycle progression and failure to disassemble the cilium is associated with defects in S-phase (7,15–17) and mitotic progression (18–21). Thus the primary cilium is not only dependent on cell cycle cues for assembly and disassembly, but also the state of the cilium has direct consequences on cell cycle progression (19,21,22).
Ciliary disassembly typically occurs in S or G2. Several cell cycle-associated regulators of ciliary disassembly have been identified (23–26). Notably, mitotic kinase Aurora A (and its homolog CALK in Chlamydomonas) activates a phosphorylation cascade required for ciliary disassembly and mitotic progression (19,21,27–32). A proposed mechanism is that the downstream target of Aurora A, the microtubule deacetylase HDAC6, deacetylates stable microtubules of the axoneme, thus destabilizing the axoneme and preparing it for subsequent disassembly, perhaps involving the IFT machinery or microtubule depolymerizing kinesins (30,33,34). Although there is little evidence for this mechanism in mammalian cells, it is consistent with a mechanism for flagellar disassembly in the unicellular green alga Chlamydomonas, in which cells gradually and simultaneously resorb their two flagella, defined as internalization of the axoneme and retention of the majority of the ciliary membrane (28,29,35–38).
Alternatively, cilia could be removed very rapidly by extracellular shedding, defined as concurrent release of membrane and axoneme from the cell body, which has also been referred to as ciliary excision, deciliation, deflagellation, or flagellar autotomy (39,40). Ciliary shedding has been observed in several ciliated protozoan species, particularly Chlamydomonas (40). Mammalian primary cilia have been observed to undergo pharmacologically-induced shedding (41), and release of a small distal portion of the primary ciliary membrane has been reported in several contexts (42–44), but the relevance of either of these events to whole cilium loss is unknown.
Here, we examined ciliary disassembly dynamics in mouse IMCD3 (inner medullary collecting duct) cells by high resolution, live-cell 3D confocal microscopy. Ciliary disassembly behaviors were remarkably dynamic and heterogeneous, and occurred on time scales ranging from seconds to hours, which we separated into 3 categories (gradual, instant, and combined) of which the fastest (101 μm/min; instant) comprised >80% of events and involved cilium shedding. We showed that katanin and intracellular calcium, two candidate regulators of ciliary shedding, independently, but not synergistically, negatively regulated cilia disassembly.
Results
Ciliary structures in cells undergoing serum-induced cilium disassembly
We manipulated serum level in mammalian cell culture medium to synchronize the presence of primary cilia in IMCD3 cells (21,45) expressing a fluorescent ciliary membrane marker (IMCD3-SSTR3::GFP) (46). The axoneme in these cells was visualized by immunostaining for acetylated tubulin (acTub), and the basal body by staining for pericentrin (PCNT) (Fig. 1A, S1A). We found that 60 +/− 9.09% of serum starved cells at 0 hour and 6 hours were ciliated after serum starvation. Re-introduction of serum (serum stimulation) over a 6 hour period resulted in a steady loss of cilia to a level of 30 +/− 0.4% that was similar to that in asynchronously cycling cells at a comparable cell density (Fig. 1B). There was also a decrease in mean ciliary length (Fig. S1C). As validation of serum-induced ciliary disassembly, cells were treated in parallel with 2 tubacin an inhibitor of HDAC6, a microtubule deacetylase required for ciliary disassembly (21,47,48), or the carrier DMSO as a control. As expected, tubacin treatment prevented the decrease in the percent of ciliated cells (Fig. 1B) and cilia length (Fig. S1C). Mitotic cells (identified by condensed nuclei mitotic spindles, separating nuclei in telophase, and cells separated by a midbody) accumulated following serum stimulation (Fig. S1); this is consistent with an inverse relationship between the presence of a primary cilium and the mitotic status (49–51). As expected, this wave of mitosis was inhibited or delayed in tubacin-treated cells (Fig. S1B).
We identified three types of cilium structures consistent with being disassembly intermediates in these fixed samples (Fig. 1C-G): 1) a discontinuous axoneme, marked by a gap in acTub staining (“Discontinuous acTub”, Fig. 1C); 2) a ciliary stub, characterized by a short (<1μm) cilium positive for both acTub and SSTR3 membrane fluorescence (“acTub+ SSTR3+ Stub”, Fig. 1D); and 3) an axoneme lacking SSTR3 fluorescence, marked by linear acTub fluorescence (>1μm) adjacent to a PCNT-labeled basal body (“acTub+ SSTR3−”, Fig. 1E). These structures were rare in serum starved cells, but at 2–3 hrs post-stimulation they comprised the majority of detectable cilia (Fig. 1F). The enrichment of these structures in post-serum stimulated cells was inhibited in the presence of tubacin (Fig. 1G), indicating that these structures are most likely representative of disassembling cilia.
The precise disassembly mechanisms represented by these different structures cannot be determined from static images; however, they illustrate that ciliary disassembly may be unexpectedly complex, and several distinct disassembly behaviors may occur within a single cell type.
Ciliary disassembly dynamics are heterogeneous and favor instantaneous cilia loss
To observe cilium disassembly directly, we generated IMCD3 cells stably co-expressing markers of the ciliary membrane (SSTR3-GFP) and basal body (mCherry-PACT). Live cells were imaged immediately following serum stimulation with full confocal stacks acquired continuously at 45–90 sec intervals for 6–12 hours on a laser scanning confocal microscope. Imaging of cilia in serum starved cells with cilia that were not disassembling (Fig. 2A.1) revealed several intrinsic features of ciliary dynamics. First, cilia underwent spontaneous length fluctuations on short-term time scales (up to ~1.5 μm between consecutive time points; Fig. S3A). Second, cilium length on average decreased slightly over the 12 hour imaging period, with a mean rate of ~0.005 μm/min (n=10, Fig. S3A), which may reflect individual cell behavior or response to our long-term fluorescence imaging regime. These features were taken into account in our analysis of cilium disassembly.
Video sequences of cells in which cilia disassembled during the course of observation, revealed a striking dynamic range of disassembly behaviors. Further analysis showed that these behaviors can be grouped into three categories, which we will refer to as gradual, instant, and combined; these are illustrated in the individual examples shown in Fig. 2A and Movies S1–4. We define these categories as follows: gradual - cilium length reduction over multiple consecutive time points ultimately leading to terminal cilium loss (e.g., Fig. 2A.2, rate of 0.016 μm/min); instant - a single discrete event of cilium loss within a single imaging frame, i.e. 30–90 sec (e.g., Fig. 2A.3, approximate minimum rate of 4.72 μm/min); and combined – a period of gradual disassembly directly followed by an instant loss event (e.g., Fig. 2A.4, gradual phase with a rate of 0.029 μm/min, followed by very rapid loss within 46 sec with an approximate minimum rate of 5.69 μm/min). We note that it was usually not possible to visualize the nature of the loss event in the instant cases due to their rapid rate and constraints on the time resolution of long time-scale live imaging.
Our initial characterization of the cilium disassembly categories relied on visual inspection of video sequences (Fig S2). To achieve a more objective analysis of disassembly, we developed a Matlab algorithm that normalized ciliary length fluctuations of individual cilia to controls, reliably identified a disassembly start point, and assigned each disassembly behavior to one of the three categories above, defined in the algorithm as follows: gradual if ciliary length just before complete disassembly (Lfinal-1) was reduced compared to the length at the start of disassembly (Lstart); instant if Lfinal-1 was greater than 1.5 μm, below which length measurements were unreliable due to possible measurement error (Fig. S3A); combined if criteria for both gradual and instant were met in the same ciliary disassembly event (Fig. S3B, Materials & Methods).
To assess overall ciliary behaviors of each category, disassembly curves were normalized by time (normalized to 1000 arbitrary units) and ciliary length (normalized to the maximum length of each cilium), and averaged (Fig. 2C-E). The averaged curve of the gradual category shows early, initial ciliary shortening with event start points (circles) distributed along the curve, followed by a period of consistent length reduction in the last ~150 normalized time units as the slope of the curve increased (Fig. 2C). The averaged curve of the instant category appears nearly horizontal until the last point, when the curve drops precipitously in a single time point; disassembly start points were nearly all clustered in the last ~10 time units (Fig. 2D). The averaged curve of the combined category was intermediate between the gradual and instant plots, with a slight downward slope of gradual dynamics followed by a rapid loss of cilia length, with start points distributed along the curve (Fig. 2E).
The resulting disassembly rates from our Matlab analysis were in the range of 10−3 – 101 μm/min (Fig. 2F). The fastest rate was for instant disassembly and was in the range of 101 μm/min. For the combined category, separate disassembly rates were calculated from the gradual and instant stages of cilia loss. Interestingly, the rate of the first gradual step of combined disassembly (0.083 μm/min ± 0.259) was not significantly different from that of the gradual-only rate (0.079 μm/min ± 0.106), and the second (instant) step (3.414 μm/min ± 2.098) was comparable to the instant-only rate (3.882 μm/min ± 1.802) (Fig. 2F). However, the majority of ciliary length in the combined category (72.6 ± 4.5%) was lost during the instant stage. These results indicate that the combined disassembly behavior likely represents both gradual and instant mechanisms within the same cilium, rather than a separate mechanism accounting for the biphasic dynamics.
Finally, we asked whether there was a correlation between the initiation of ciliary disassembly after serum stimulation and disassembly behavior. Disassembly start and end times were plotted as a histogram (Fig. S4). Gradual disassembly events most frequently initiated in the first two hours of serum stimulation (Fig. S4A), although the sample size was low due to the relative rarity of this event (n=11). Instant disassembly events increased progressively over the first three hours of stimulation, and appeared to be more widely distributed throughout the time course (Fig. S4B). Interestingly, combined category disassembly start times, marking the onset of the initial gradual stage, were most frequent in the first two hours after stimulation. Combined end points, which occurred with instant dynamics, progressively increased over the first three hours, in agreement with the instant category (Fig. S4C). These results further illustrate that combined disassembly behaviors share features of gradual and instant disassembly behaviors.
Taken together, our analysis identified three categories for primary cilia disassembly behavior in IMCD3 cells: gradual, instant, and combined. Importantly, the instant and combined categories jointly account for 83.1% (n=70) of all ciliary disassembly events (Fig. 2G), and the instant step in the combined category accounted for approximately 75% of ciliary length loss.
Instant cilia loss dynamics are consistent with extracellular shedding
Our results show that in mammalian cells the final stages of cilia disassembly and loss are on the time scale of seconds, which is several orders of magnitude more rapid than expected for resorption mechanisms as reported in Chlamydomonas. However, much more rapid cilium retraction into the cell has been reported in other unicellular eukaryotes, such as the chytrid fungi (52). To better determine the mechanism of cilium disassembly in mammalian cells we sought to identify sequences in which the loss event could be visualized.
In a rare instance, we observed direct shedding of ciliary membrane from the cell surface (Fig. 3A, Movie S5). In addition, ciliary shedding induced by treatment with dibucaine had similar qualitative features and dynamic profiles to serum-induced instant disassembly (Fig. S5) (40,41,53). The rapid nature of complete ciliary detachment in under 54 secs, and the diffusion of the shed remnant(s) away from the site of origin is consistent with our observations of the disappearance of the ciliary membrane with instant dynamics in 83.1% of disassembly events (Fig. 2G).
The shed cilia in Fig. 3A have a fragmented appearance. We interpret this as being the result of the free ciliary fragment moving at a faster rate than the rate of z-stack acquisition, rather than sequential loss of small segments of the cilium. Because the shed cilium travels some distance in the time between individual z-slice acquisitions, the apparent location of an object in that slice will have shifted. As a result, the final imaged object appears distorted – artificially elongated or with the appearance of separated fragments. We note that the membrane segments in such sequences were usually visualized as a group that moved together, rather than dispersing independently, consistent with the interpretation that they are all part of the same structure imaged at different times.
We asked whether the cilium was released with or without the axoneme. Excision of portions of the ciliary membrane alone has been previously reported (42–44). To test this, IMCD3-SSTR3::GFP cells were transiently transfected with mCherry-α-tubulin, stimulated with serum, and imaged at 30 sec intervals. Tubulin was observed in shed ciliary fragments (Fig. 3B, Movie S8), suggesting that the axoneme is shed together with the ciliary membrane and distinguishing this event from previously described ciliary membrane-fragment or ectosome release (42–44).
Validation of ciliary shedding by isolation of ciliary fragments from culture media
Observation of discrete shedding of the entire visible ciliary membrane supported the hypothesis that this behavior underlies instant ciliary disassembly dynamics (Fig 3A). Given the limitations of imaging described above, it was important to demonstrate the form in which the cilium is shed. We sought to detect ciliary fragments in the culture medium, predicted to be present based on the hypothesis of shedding as the mechanistic basis for instant disassembly.
Previously described methods of cilia isolation from mammalian cells have required artificial deciliation to synchronously release sufficient amounts of ciliary material for detection (41,54). To detect ciliary fragments released spontaneously from serum-stimulated cells we developed two novel methods (Fig. 3). First, immune-capture of cilia allowed direct visualization of the size and shape of unperturbed, shed ciliary fragments. In brief, culture medium from serum-stimulated cells was incubated on glass-bottomed imaging dishes coated with an antibody specific for the extracellular N-terminus of SSTR3 (Fig. S6A). Cilia expressing SSTR3-GFP bound the antibody (Fig. S6B), and were imaged, unfixed, by fluorescence (Fig. 3C) or after fixation, by electron microscopy (Fig 3E). A representative image of an antibody-captured, intact cilium is shown from IMCD3-SSTR3::GFP cells transfected with mCherry-α-tubulin to mark the axoneme (Fig. 3C). The jagged appearance of cilia in these samples is due to thermal motion in surrounding aqueous solution during stack acquisition. Quantification showed a 4-fold increase in cilia captured from medium from serum-stimulated cells compared to control medium from serum-starved cells. As expected, pre-treating serum-stimulated cells with tubacin resulted in a decrease in the amount of captured cilia to levels in the control (Fig. 3D).
However, the cilia immune-capture method was limited by low amounts of material and instability of antibody-bound cilia, which made analysis by immunofluorescence or biochemistry difficult. Therefore, we developed a filter-spin cilia isolation method to concentrate large volumes of medium containing ciliary fragments (Fig. 3F&G, S6B). 40–80 mL of serum-starved or - stimulated culture media were subjected to a combination of filtration and centrifugation steps to concentrate ciliary material approximately 500-fold. As a positive control, serum-starved cells were scraped and treated with high-calcium deciliation buffer (55) to induce artificial shedding of cilia (see Materials and Methods). Immunostaining for ciliary markers (IFT88 and α-tubulin) demonstrated increased abundance of ciliary structures in the concentrated material, many with the dimensions expected of shed primary cilia (Fig. 3F). Western blotting for IFT88, α-tubulin, and acetylated tubulin further confirmed detectable levels of cilia-specific proteins in the concentrated medium (Fig. 3G).
Together, immune-capture and filter-spin concentration methods for ciliary isolation demonstrated the presence and correct dimensions of intact cilia containing ciliary membrane (SSTR3), axonemal tubulin, and other specific ciliary markers that were shed from serum-stimulated cells. The consistent detection of full-length, tubulin-containing cilia by two independent methods indicates that ciliary shedding is a physiological behavior of IMCD3 cells, and is a likely explanation for the most prevalent, instant ciliary disassembly behavior.
p60 katanin activity regulates ciliary disassembly behaviors
We next examined the roles of potential regulators of ciliary shedding in the disassembly behaviors that we had observed. In Chlamydomonas, the deflagellation-incompetent mutant Fa1p carries a mutation in a gene homologous to katanin, a conserved microtubule-severing AAA ATPase with roles in mitotic spindle formation and function, and axon extension in mammalian cells (56–58). In Chlamydomonas, katanin directly triggers axoneme severing in vitro and localizes to the site of axoneme breakage at the transition zone (39,59,60). Katanin overexpression also induced ciliary disassembly in Tetrahymena (61).
In order to determine whether katanin activity contributed to ciliary disassembly by shedding in mammalian cells, we overexpressed the catalytic domain of mouse katanin, p60 (KATNA1) in IMCD3-SSTR3::GFP cells. We imaged tRFP fluorescence for the transfected p60, and used a p60-specific antibody for total katanin (including endogenous protein) in IMCD3-SSTR3::GFP-turboRFP(tRFP) control and IMCD3-SSTR3::GFP-tRFP::p60 cells. Both endogenous p60 and tRFP-p60, but not tRFP alone, were located at mitotic spindles and diffusely in the cytoplasm. In some cells, a punctum of katanin fluorescence was clearly detected at the basal body of cilia (Fig. 4A). The level of cytoplasmic acetylated tubulin was reduced significantly in serum-starved and -stimulated (2 hr) tRFP-p60 expressing cells compared to the control (Fig. S6A-B), indicating that over-expressed katanin was active and induced increased severing and destabilization of cytoplasmic microtubules (61,62).
We assessed the effect of p60 overexpression on ciliary abundance, length, and disassembly behavior on a population level. Fixed, serum-starved tRFP- and tRFP-p60-expressing cells displayed similar levels of ciliation and responded similarly to 6-hour serum stimulation (Fig. S6C). However, ciliary length was significantly reduced in tRFP-p60 cells, as measured from confocal z-stacks of live cells (Fig. 4B). Tubacin and cytochalasin D treatments inhibited ciliary disassembly in both tRFP and tRFP-p60 cells (Fig. S6D). These results indicate that p60 activity did not impair overall ciliogenesis or spontaneously induce ciliary disassembly, but reduced ciliary length.
Next, we asked whether tRFP-p60 overexpression affected ciliary disassembly dynamics. Ciliary disassembly events were observed and analyzed as described above (see Fig. 2). In tRFP expressing cells, combined disassembly was the largest category of disassembly events (53.3%, n=61). In tRFP-p60 cells, gradual disassembly was virtually eliminated, while the frequency of instant disassembly increased ~33% (n=50). Cumulative average disassembly curves normalized to disassembly time and cilium length further illustrated a difference in ciliary disassembly dynamics (Fig. 4D-G). Due to the low frequency of gradual disassembly events, only combined and instant behaviors are discussed further. In both combined and instant disassembly categories, the cumulative curves from tRFP expressing control cells (Fig. 4D&F) have a steeper slope than the equivalent plots from p60 overexpressing cells (Fig. 4E&G). These data demonstrate that in tRFP-p60 cells ciliary disassembly by gradual dynamics (either gradual-alone or combined events) was generally reduced, and instant disassembly was more prevalent.
Intracellular calcium does not act synergistically with p60 to promote ciliary shedding
High extracellular calcium and drug-induced increases in intracellular calcium levels have been used to artificially force ciliary shedding in ciliates, flagellates, and mammalian cells (40,41,55,63). However, mechanisms underlying this effect have not been elucidated. Calcium-calmodulin signaling is upstream of AurA, a key activator of ciliary disassembly (20), indicating a role for changes in cytoplasmic calcium levels in ciliary disassembly. We tested whether or not increased intracellular calcium levels and katanin worked synergistically to promote ciliary shedding.
To examine the roles of intracellular calcium levels and p60 overexpression in cilia disassembly, we used well-known small molecule modulators of intracellular calcium levels (dibucaine (40,41,64,65), ionomycin (66), and thapsigargin (41,67) increase cytoplasmic calcium levels; BAPTA-AM, a calcium chelating agent, reduces calcium levels (68)) on tRFP- and tRFP-p60-overexpressing IMCD3 cells, and ciliation rates and length were quantified (Fig. 5).
tRFP and tRFP-p60 expressing cells were serum-starved, pre-treated with dibucaine, ionomycin or thapsigargin, fixed, and the percentage of cells that were ciliated was measured. Cilia counts for each treatment were normalized to levels in DMSO-treated cells to give a relative change in ciliation. Both dibucaine and ionomycin caused a significant reduction in the abundance and length of cilia in both tRFP- and tRFP-p60-expresssing cells (Fig. 5AB). These results indicate that increased cytoplasmic calcium levels, either through release of intracellular membrane-bound calcium (dibucaine, (64,65)) or influx of extracellular calcium (ionomycin, (66)), had similar effects in negatively regulating primary cilia, and that tRFP-p60 overexpression did not affect this response. In contrast, 5 μM thapsigargin, which releases intracellular calcium stores (67,69), reduced ciliary abundance but not length in tRFP expressing cells, and increased the abundance and length of cilia in tRFP-p60 expressing cells (Fig. 5CD).
To reduce the level of cytoplasmic calcium, serum-starved cells were pre-treated with BAPTA-AM for 30 min. Consistent with previously published results (20), the percent of ciliated cells after 2 hr and 6 hr serum stimulation was reduced in BAPTA-but not DMSO-treated tRFP expressing cells, demonstrating that ciliary disassembly was inhibited by intracellular calcium chelation. In contrast, tRFP-p60 expressing cells had levels of ciliary disassembly that were similar in DMSO and BAPTA-AM treated conditions (Fig. 5E). In addition, BAPTA-AM treatment caused a significant reduction in overall ciliary length in tRFP-p60, but not in tRFP expressing cells (Fig. 5F). Taken together, these results indicate that intracellular calcium and katanin do not synergistically promote ciliary loss. The negative relationship between intracellular calcium levels and cilia disassembly may be dependent on the source or concentration of intracellular calcium, and p60 overexpression alters this relationship.
Discussion
We examined cilium disassembly in mammalian cells, using live-cell single-cilium analysis, revealing highly heterogeneous rates of cilium loss, spanning approximately three orders of magnitude, from hours to seconds. These behaviors fell into three categories – gradual disassembly (10−3 – 10−1 μm/min), instant disassembly (101 μm/min), and combined disassembly, consisting of consecutive gradual and instant dynamic phases. We conclude from these observations that there may be different mechanisms by which a cilium is disassembled within a single cell type, and even within a single cilium (in the case of combined disassembly). We propose that gradual disassembly is explained by disassembly of the axoneme and ciliary resorption, and instant disassembly is best explained by ciliary shedding. Interestingly, the shedding and resorption behaviors described in Chlamydomonas, chytrid fungi, and other organisms, are conserved in very distantly-related mammalian cells.
The co-existence of resorption and shedding in combined disassembly is intriguing, as it indicates differential regulation of the distal and proximal portions of the cilium leading to different methods of disassembly in each. A similar phenomenon has been described in Chlamydomonas, in which the initial resorption of the bulk of the flagellum was followed by severing the remainder from the basal body and release of a small particle into the surrounding medium (36). The wide range of disassembly rates in the gradual category that we observed in IMCD3 cells indicates that there are several resorption mechanisms, or one mechanism with highly tunable dynamics. Thus, a combined resorption-severing behavior may be conserved between Chlamydomonas and mammalian cells.
What could contribute to differential regulation of distal and proximal portions of the cilium? One study showed that that different kinases and downstream effectors control the disassembly of the distal and proximal regions of the flagellum, but both contribute to resorption (29). It is also becoming apparent that structural features of the axoneme may contribute to the regulation of different regions of the cilium, such as the transition from doublet to singlet microtubules or the distribution of microtubule post-translational modifications that are non-uniformly distributed in the axoneme (70–72). Additionally, there may be a length- or time-dependent switch or signal that activates a new mechanism once the cilium has been partially disassembled. The conditions and factors underlying the decision to undergo one type of disassembly over another are likely complex and will require further study.
Taking instant and combined disassembly categories together (because the last stage of combined disassembly is an instant loss event), instant dynamics accounted for the majority of the ciliary disassembly events observed in IMCD3 mammalian cells. Direct observation of tubulin and the membrane marker SSTR3 during ciliary loss from the cell surface indicated extracellular shedding of cilia. This was further supported with two independent methods that isolated ciliary fragments shed into the medium. The presence of ciliary material in serum starvation media was unexpected, but repeatable, and due to two major causes – 1) ciliary shedding is a means for ciliary disassembly, but does not exclusively occur in disassembling cilia, as we have observed shedding and immediate ciliary regrowth in serum starved cells (data not shown), and 2) cilia yields from the filter-spin concentration method are not reliably quantitative due to the potential for sample loss at several steps.
Regardless of these caveats, the morphology and composition of captured cilia confirmed 3 major points regarding mechanisms of ciliary shedding: 1) cilia are shed as intact structures; 2) the fragmented appearance of shed cilia in live cell imaging are likely the result of confocal imaging of a highly dynamic process, rather than cilium fragmentation; and 3) shed cilia contain tubulin, implying that the axoneme is severed and shed along with the ciliary membrane. These results are also consistent with the observation of discontinuous axonemes, and the frequent observation of instant disassembly in which the entire ciliary membrane is shed cleanly from the basal body in response to serum stimulation or dibucaine. Together, these observations strongly support the hypothesis that highly prevalent instant disassembly is due to ciliary shedding, and distinguishes our results from previous studies that described the release of ectosomes, apical abscission, and decapitation of the ciliary membrane alone (42–44).
A limitation of our live cell imaging of ciliary disassembly was the high fluorescence background in the cytoplasm, which meant that we could not follow the fate of the axoneme and membrane during ciliary resorption. Nevertheless, in fixed cells we identified several novel non-canonical ciliary structures in serum-stimulated cells that might be ciliary disassembly intermediates, which might provide insight into mechanisms for ciliary shedding and resorption, for example: 1) discontinuous acetylated tubulin staining indicates a break in the axoneme, and the accompanying constriction at that site in SSTR3 membrane fluorescence indicates membrane pinching that could portend ciliary severing at that site prior to shedding; 2) a ciliary stub positive for acetylated tubulin and SSTR3-GFP could represent a remnant of a shed or resorbed cilium close to the cell surface (73), although at this resolution it cannot be determined whether the ciliary stub is on the cell surface or in the cytoplasm; and, 3) acetylated tubulin fluorescence without corresponding SSTR3 membrane fluorescence indicates a resorbed axoneme-basal body complex in the cytoplasm, as has been observed previously (74), in which the membrane was released or incorporated into the plasma membrane. While these interpretations are speculative, due to the nature of static representations of a dynamic process and the markers of ciliary structures, the relative lack of these ciliary structures in starved and tubacin-treated conditions indicates that they may be representative of the ciliary disassembly process.
We assessed the roles of the microtubule severing enzyme katanin and intracellular calcium in regulating ciliary shedding. Katanin mediates axoneme severing and ciliary shedding in Chlamydomonas and Tetrahymena (39,60,61,75), and high intracellular calcium levels trigger ciliary shedding in Chlamydomonas and mammalian cells (39–41). Overexpression of the katanin catalytic domain p60 reduced ciliary length, and nearly eliminated the gradual category of ciliary disassembly, which was compensated for by an increase in the instant disassembly category. Therefore, upregulation of katanin activity would likely promote instant ciliary disassembly behavior, and, by extension, ciliary shedding. Interestingly, we did not find that p60 overexpression affected general ciliation, in contrast to previous work showing a ciliogenesis defect in response to overexpression of the related proteins katanin-like 2 (KATNAL2) (76) and fidgetin-like 1 (FIGL1) (77). This may be due to differences in protein functions or levels of protein expression, or criteria for categorizing short versus absent cilia. Alternatively, p60-overexpression in serum-starved cells may cause ciliary severing at sites other than the ciliary base (Fig. 1C, 3B), or shedding from the base might be rapidly followed by cilium regrowth in quiescent cells (M. Mirvis, unpublished results), either of which could result in an apparent decrease in the average cilium length in a cell population.
Based on previous work in Chlamydomonas showing that both katanin and raised intracellular calcium levels spontaneously induce ciliary breakage and severing in vivo and in vitro (39), we tested whether they act synergistically in promoting ciliary shedding and overall disassembly in IMCD3 cells. Consistent with previously published results, we found that addition of drugs that raise intracellular calcium levels reduced ciliary number and length, while chelating intracellular calcium inhibited ciliary disassembly in control cells (20,41). Unexpectedly, overexpression of p60 had opposing effects – addition of thapsigargin increased the number and length of cilia, while BAPTA-AM shortened cilia and failed to inhibit serum-induced ciliary loss. Thus, intracellular calcium levels and katanin do not act synergistically or additively to promote ciliary shedding. Little is known regarding a direct relationship between calcium and katanin, although one study showed that p60 has several calcium binding sites, and that calcium binding inhibited p60 severing activity (78). Interestingly, high calcium levels have been shown to affect induce primary cilium bending by altering axoneme microtubule morphology (79).
Our work raises many new questions regarding the mechanisms and dynamics of primary ciliary disassembly that will need to be addressed in future studies. Why do multiple mechanisms for ciliary disassembly exist? Could ciliary resorption and shedding have specific advantages for cell cycle regulation or signaling? How is the ciliary membrane separated from the cell during ciliary shedding? Recently published work identified an actin-dependent phosphoinositide-based pathway underlying ciliary decapitation (43), and ciliary disassembly activator Aurora A may act upstream of this pathway (45). Future work may focus on whether the same mechanism is responsible for membrane detachment at the ciliary base. What is the precise mechanism of axoneme severing by katanin? Future studies using immunogold electron microscopy could determine whether katanin is localized specifically to the transition zone in mammalian cells, as shown in Chlamydomonas (59). Furthermore, answering how katanin carries out complete, localized, coordinated disruption of the axoneme despite the complex microtubule structure of the axoneme may provide novel insights into general mechanisms for microtubule severing.
Materials & Methods
Cell Culture
IMCD3 cells were grown in DMEM-F12 medium with 10% fetal bovine serum and 1% penicillin-streptomycin-kanamycin antibiotic cocktail. Cells were passaged every 2–3 days at 1:10-1:20 dilution. Cells were tested for mycoplasma with Sigma LookOut Mycoplasma PCR Detection Kit (Cat#MP0035) as directed by the manufacturer, and incidences of mycoplasma contamination was treated with Mycoplasma Removal Agent (MP Biomedicals, #093050044). Following decontamination, experiments potentially affected were repeated at least three times to determine any difference in results. No significant differences were observed.
Serum Starvation and Stimulation
Cells were seeded in 24- or 6-well dishes with glass coverslips for imaging following fixation, or 35 mm glass-bottomed MatTek dishes (#P35G-0-10-C) for live imaging. 24-well dishes were seeded at 1.5×104 cells and 6-well and 35 mm MatTek dishes were seeded at 1-1.5×105, to achieve 50–70% confluence next day. For serum-starvation, cells were washed once with 0.2% DMEM-F12 + PSK, then grown in 0.2% DMEM-F12 + PSK for 24 hr. Serum stimulation was by either re-addition of FBS directly to dishes to 10% final concentration, or replacement with 10% FBS DMEM-F12.
Antibodies
The following antibodies and dilutions were used. Acetylated tubulin mouse monoclonal 6-11B-1 (1:1000 for IF & WB) (Sigma-Aldrich Cat# T7451); Pericentrin rabbit polyclonal Poly19237 (1:500 for IF) (Covance Cat #PRB-432C, now BioLegend); Arl13b rabbit polyclonal (1:250-1:500 for IF) (Proteintech Cat# 17711-1-AP); N19-SSTR3 antibody (rabbit polyclonal) (Santa Cruz Cat #sc-11610, discontinued); IFT88 rabbit (1:500 for IF & WB) (GeneTex, Cat#79169); α-tubulin YL1/2 (1:1000 for IF & WB), (ThermoFisher #MA1-080017); alpha-tubulin DM1a (1:1000 for IF & WB) (ThermoFisher #62204); rabbit monoclonal anti-p60 EPR5071, (1:250 IF), (Abcam Cat# ab111881); rabbit polyclonal anti-KATNAl, (1:100-250 IF) (Proteintech Cat#17560-1-AP). Anti-rabbit GFP (1:250) (Life Technologies, #A11122); Anti-mouse GFP (1:1000) (Roche, #11063100). Secondary antibodies used were: Anti-mouse Rhodamine (1:1000) (Jackson ImmunoResearch, #715-295-150), Anti-rabbit FITC (1:1000) (Jackson ImmunoResearch, #111-095-003), Anti-rabbit Alexa647 (1:200) (Life Technologies, #A21245), Anti-mouse Alexa647 (1:200) (Life Technologies, #A21236), Hoescht (1:1000-2000) (Molecular Probes, #H-3570).
Chemicals
Tubacin (Sigma-Aldrich, #SML0065) at 2 μM in DMSO; dibucaine hydrochloride, (Sigma-Aldrich #285552) at 190 μM in DMSO. The following were used at 1μM in DMSO: Cytochalasin D (#PHZ1063), Thapsigargin, (Sigma-Aldrich #T9033). BAPTA-AM (Sigma-Aldrich #A1076). Ionomycin 10 mM stock was a gift from the Lewis laboratory, Stanford Univ.
Generation of stable cell lines
IMCD3-SSTR3::GFP-mCherry::PACT: mCherry::PACT was cloned from a pLV plasmid (pTS3488, created by multi-site Gateway cloning by Christian Hoerner) onto pLV-Puro-EF1a construct using Gibson cloning. Lentivirus with the cloned construct was generated in HEK293T and used to infect IMCD3-SSTR3::GFP (gift from Nachury laboratory, (46)) under selection with 800 mg/mL puromycin for 4–5 days. Infected cells were FACS-sorted into polyclonal populations by mCherry fluorescence intensity, and a pool of low-expressing cells were selected to prevent over-expression phenotypes of a centrosomal protein.
Katanin expression constructs: Mammalian expression constructs for turboRFP and turboRFP::p60 (p60 domain of mouse katanin) were designed and ordered from VectorBuilder. All constructs were amplified by transformation in DH5a and maxi-prep (Qiagen #12165). IMCD3-SSTR3::GFP cells were transfected with each construct with ThermoFisher Lipofectamine 3000 according to manufacturer’s protocol (#L3000015). The next day, cells were subjected to G418 selection (800ng/μL for 5–6 days). Cells were sorted by FACS into low-, medium-, and high-expressing pools, and maintained in DMEM-F12 10% FBS + PSK and 250 ng/μL G418 to maintain transgene expression.
Transient transfection
mCherry-α-tubulin mammalian expression construct was a gift from Angela Barth, Stanford Univ., and transfected into IMCD3-SSTR3::GFP cells. Transfections were performed using Lipofectamine 3000 transfection reagent according to manufacturer’s protocol.
Immunofluorescence microscopy
Generally, fixation for immunofluorescence microscopy was done with 100% methanol for 5 minutes at −20°C, followed by washes with 0.1% Triton X-100 in PBS at room temperature for 2 minutes, and 3 PBS washes. Samples were blocked for 1 hr at RT°C or overnight at 4°C in 2% BSA, 1% goat serum, 75mM NaN3. Antibodies were diluted to indicated concentrations in blocking buffer. Primary antibody incubations were performed for 1 hr at RT°C or overnight at 4°C. Secondary antibody incubations were performed for 1–2 hr at RT°C. Following each antibody incubation, samples were washed 3 times in PBS + 0.05% Tween-20 for 5 mins each at RT°C.
Images were acquired with a Zeiss Axiovert 200 inverted epifluorescence microscope and a 63x objective, or a Leica SP8 scanning laser confocal microscope with LASX Software, using mercury or argon lamps with white light laser excitation, and a 63x 1.4 NA oil objective. Exposure times were constant during each experiment. For imaging of serum-starved and serum-stimulated cells, fields of view were selected based on DAPI staining by two critieria: 1) to select for moderate cell density, in order to avoid effects of high density on cell cycle and ciliation; and 2) to eliminate bias in % cilia quantifications from scanning by ciliary markers.
Live-cell confocal microscopy
Cells were cultured in glass-bottomed Mattek dishes and imaged in DMEM-F12 media with 15mM HEPES without phenol red. Movies were acquired 4–12 hr after serum stimulation with a Leica SP8 scanning laser confocal microscope using 0.5 μm z-slices, 30–90 sec intervals, autofocus, in a 37°C incubator, and red and green channels were acquired simultaneously. The video file was saved as .lif from LASX software and opened in Imaris x64 8.0.2 as a 3D render for analysis of cilia disassembly dynamics and basal body positioning.
Data analysis
Cilia counts and length measurements were performed either manually in Fiji or Imaris x64 8.0.2 and 9.2.1, or through semi-automated detection in Imaris. Manual analysis involved detecting ciliary membrane, marked by an enrichment of SSTR3::GFP above background threshold, that were adjacent to a centriole (mCherry::PACT in dual-fluorescent cells or pericentrin immunofluorescence in single-(SSTR3::GFP-expressing) or non-fluorescent cells), to distinguish from accumulations of SSTR3+ membrane elsewhere in the cell. Manual length measurements in Fiji were made with the line function, and in Imaris with the Measurement tool. Generally, single z-plane images were analyzed in Fiji or Imaris, while confocal z-stacks were analyzed in Imaris which allowed more accurate length measurement due to the 3D render (Surpass) capability. When possible, length measurements in confocal images were semi-automated in Imaris using the Surfaces function to create an artificial object encompassing the ciliary membrane, and exporting Bounding Box data as a proxy for length (the longest dimension of the object).
For live cell serum stimulation experiments, movies were visually scanned in Imaris for examples of disassembling cilia. Images of each disassembling cilium were cropped by time (from t0 to several mins after complete loss), and position (restricted to area of occupancy during the that time window), and then saved in a separate file. To generate ciliary length curves, the ciliary membrane was isolated as an artificial object using the Surface function. When possible, the object was automatically tracked over consecutive time points with length data generated at each time point. In cases where automatic tracking was not possible due to low signal-to-noise of ciliary membrane fluorescence, measurements were taken manually at 15–30 minute intervals until the initiation of ciliary disassembly, and at each time point during the disassembly event.
Matlab
Raw length measurement data from disassembling cilia movies were compiled in to an Excel spreadsheet. A Matlab algorithm imported the data, performed smoothing and calculations of cilium start point, and generated an output file containing disassembly rates, start and end time, start and end length, and proportion length lost per disassembly stage. Algorithm strategy is described in the text and Supplement.
Cilia Isolation
Cell culture: clones of IMCD3 cells, either an unsorted stably expressing GFP-SSTR3 or FACS sorted for medium expression of GFP-SSTR3, were grown on 15cm dishes at 3×106 cells/dish in DMEM/F12 with 10% FBS and antibiotics for 24 hours. The cells were washed 3x with HDF wash buffer, and media was replaced with DMEM/F12 and 0.2% FBS and antibiotics (serum-starved) for 24 hours. Then all dishes were washed 3x with HDF buffer and half received phenol-red free DMEM/F12 with 0.2% FBS (serum-starved), and the other half received phenol red free DMEM/F12 with 10% FBS (serum-stimulated) for 24 hours. Total serum starved time was 48 hours and total serum stimulated time was 24 hours.
Immune-capture Method
Preparation of antibody-immobilized imaging dishes: Glass in 35 mm glass-bottomed imaging dishes (MatTek) was functionalized by plasma cleaning at 250 moor, Low setting, 45–60 seconds. Dishes were silanized with 500 μL of 2.5% triethoxysilyl-undecanal (TESU) in 100% ethanol, covered with Parafilm and incubated at RT°C for 1hr. Dishes were washed 3x with 100% ethanol, then baked at 85°C for 3 hrs. Next, silanized dishes were treated with the following series of reagents for 1hr at RT°C unless stated otherwise, with 3 PBS washes in between steps: 1) 50mM NHS-LC-LC-biotin in water, 2) 5mg/ml neutravidin for 1hr at RT°C, 3) 300 μg/ml biotin-Protein A 4) block with 15mM D-biotin in DMSO for 30 min, 5) anti-rabbit SSTR3 N19 (extracellular N-terminus) antibody (100–200 μg/mL) at 37°C, followed by one PBS wash. These protocols adapted from Dr. Nicholas Borghi (80).
Sample preparation
Culture medium was collected and subjected to centrifugation for 10 min at 1000xg at 4°C to remove large cell debris. Samples were then kept on ice until plating on treated dishes or stored at 4°C for a maximum of 1 day. 4mL serum-stimulated or -starved medium was incubated on a treated MatTek dish overnight at 4°C, followed by 3 gentle PBS washes. Samples were then imaged directly, without fixation with a Leica SP8 confocal microscope.
SEM
Antibody-immobilized MatTek dishes incubated with serum stimulation medium were fixed for SEM in 4% PFA, 2% glutaraldehyde, and 0.1M Na cacodylate. Glass bottoms were removed, processed for imaging, and imaged with a Hitachi S-3400N VP SEM scope in the Beckman Imaging Facility, Stanford Univ.
Filter-Spin Concentration Method
Harvest of Cilia: Deciliation of starved IMCD3 cells (positive control): Serum-stimulated or -starved culture medium, or fresh culture medium (with 10% FBS, an additional control) was removed from 6 150cm dishes, combined and centrifuged at 200xg at 4°C for 5 mins in the A-4-81 rotor, Eppendorf 5810R centrifuge. Cells were washed 2x with warm PBS containing 0.4% EDTA. 10 mL was added to a MatTek dish and incubated for 10 min at 37 °C, followed by gentle up and down pipetting to remove cells from dish. An aliquot of cell suspension was removed for cell count. Cells were centrifuged at 13000 × g for 5 mins at RT°C. The cell pellet was resuspended in 5 mL ice cold deciliation buffer (55) (112 mM NaCl, 3.4 mM KCl,10 mM CaCl2, 2.4 mM NaHcO3, 2 mM HEPES, pH7.0 and a protease inhibitor tablet [Roche]). The cell suspension was incubated at 4 °C for 15 mins with rigorous end-over-end rotation, and then centrifuged at 1000 × g for 5 mins at 4 °C in an Eppendorf centrifuge. The resulting supernatant was used for biochemistry and immunostaining.
Biochemistry
Half of the supernatant material from deciliated, serum starved or -stimulated cells was centrifuged at 21,000×g for 15 min in JA25.5 rotor in Beckman Coulter Avanti J-25I centrifuge at 4°C. The supernatant was carefully removed, and pellets were resuspended in 160 pl of sample buffer (1% SDS, 10mM Tris-HCl, pH 7.5, 2mM EDTA). Samples were boiled at 95°C for 8 mins, and equal volumes were separated by 10% PAGE and transferred to PVDF. Blots were blocked (2% BSA, 1% normal donkey and goat serum in TBS, pH 7.4) for 1hr at RT°C or overnight at 4°C. Membranes were blotted with YL1/2 (1:1000), mouse acetylated-tubulin antibody (1:1000), and IFT88 rabbit antibody (1:500) in blocking buffer for 1 hr at RT°C. Blots were washed 5x with TBST. Secondary anti-rabbit, anti-mouse, or anti-rat antibodies labeled with either Alexa Fluor 680 (InVitrogen, #A21058) or IRDye800CW (Li-Cor Biosciences, #926-32213), at 1:30,000 dilution were incubated with blots for 30 min at RT. Blots were washed 5x with TBST and scanned on Licor Odyssey scanner (Li-Cor BioSciences).
Immunofluorescence
Half of the supernatant material from deciliated, serum-starved, or -starved and -stimulated cells was concentrated using a 250ml 0.2μm PES filter unit with house vacuum to reduce the volume to 2 mL, and finally a Millipore Ultrafree-MC filters (PVDF 0.2μm size #UFC30GV100) to reduce the volume to ~0.5 mL. 5 μl of concentrated supernatant was pipetted onto an acid-treated glass slide. A 22 mm acid-treated circular glass coverslip was placed on the sample, and the slide was immediately plunged into liquid nitrogen for ~5 sec. After removing the slide, the coverslip was removed and fixed in −20 °C 100% methanol for 5 mins. Immunofluorescence staining was performed as described above.
Statistics
All analyses were performed in GraphPad Prism. Statistical tests used for each analysis are indicated in the Figure legends. No explicit power analysis was used to determine sample size. All experiments were performed with at least three biological replicates, i.e. samples from independent cell culture passages. When used, technical replicates (i.e. repeats from the same cell culture passage) were averaged for each biological replicate. In brief, comparisons of mean values such as mean percent cilia across replicate experiments were compared using an unpaired t-test. Analyses of individual measurements such as cilia length were subjected to normality tests (Kolmogorov-Smirnoff, D’Agostino & Pearson, and Shapiro-Wilk). If data passed all normality tests, unpaired t-test was used, if not the Mann-Whitney U test was used. If data passed normality by some tests but not others, both types of analyses were performed. Results were similar between parametric and nonparamentric tests.
Supplementary Videos
Video S1. Primary cilium of a serum starved cell. Imaged at 90 second intervals over 12 hrs, 45fps. The cilium undergoes rapid length fluctuations and a slight overall reduction in length at 0.003 μm/min. See Fig. 2A.1.
Video S2. Primary cilium disassembling by gradual dynamics. Imaged at 90 second intervals, 45fps. Cilium disassembles with an overall rate of 0.016μm/min. See Fig. 2A.2.
Video S3. Primary cilium disassembling by instant dynamics. Imaged at 90 second intervals, 15fps. Cilium is lost in under 90 seconds, approximate minimum rate of 4.72μm/min. See Fig. 2A.3.
Video S4. Primary cilium disassembling by combined dynamics. Imaged at 46 second intervals, 25fps. Cilium undergoes gradual shortening (0.029μm/min), followed by instant loss (>5.69μm/min). See Fig. 2A.4.
Video S5. Primary cilium shedding. Imaged at 54 second intervals. Ciliary membrane is released from cell surface (27:12–29:00). See Fig. 3A.
Acknowledgements
We thank Jonathan Indig for critical assistance with developing and writing the Matlab algorithm, Fan Ye and the Max Nachury lab for the gift of IMCD3-SSTR3::GFP cells, Martijn Gloerich for assistance with lentivirus, Daniel Cohen and Caitlin Collins for technical assistance with cilia immune-capture, Lydia Joubert and the Beckman Cell Sciences Imaging Facility for assistance with SEM sample preparation and imaging. We thank Jessica Feldman, Lucy O’Brien, Jackson Liang, and members of the Nelson and Stearns laboratories for invaluable discussion of the methods and results. Research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health under award number T32GM007276 (M.M.)., R35 GM118064 to W.J.N. and R01GM121424 to T.S. The content is solely the responsibility of the authors and does not necessarily represent the official view of the National Institutes of Health.