A homeostatic clock sets daughter centriole size in flies

Abstract

Centrioles are highly structured organelles whose size is remarkably consistent within any given cell type. New centrioles are born when Polo-like kinase 4 (Plk4) recruits Ana2/STIL and Sas-6 to the side of an existing "mother" centriole. These two proteins then assemble into a cartwheel, which grows outwards to form the structural core of a new daughter. Here, we show that in early Drosophila melanogaster embryos, daughter centrioles grow at a linear rate during early S-phase and abruptly stop growing when they reach their correct size in mid- to late S-phase. Unexpectedly, the cartwheel grows from its proximal end, and Plk4 determines both the rate and period of centriole growth: the more active the centriolar Plk4, the faster centrioles grow, but the faster centriolar Plk4 is inactivated and growth ceases. Thus, Plk4 functions as a homeostatic clock, establishing an inverse relationship between growth rate and period to ensure that daughter centrioles grow to the correct size.

Figure 1.

Sas-6-GFP is incorporated irreversibly into the growing daughter centriole. (A and B) Micrographs show a 3D-SIM-FRAP analysis of Sas-6-GFP (A) or Ana2-GFP (B) dynamics. A 3D-SIM image of a centriole pair was acquired in early S-phase (Pre-bleach); the centrioles were then photobleached (Bleach), and a post-bleach image was acquired 5–10 min later. (C and D) The same image acquisition protocol was followed in embryos that simultaneously express the mother centriole marker Asl-mCherry along with either Sas-6-GFP (C) or Ana2-GFP (D; note that bleaching the Sas-6- or Ana2-GFP, as an unintended consequence, also bleaches the Asl-mCherry fluorescence). Asl-mCherry fluorescence rapidly recovers at the mother centriole, as reported previously (Novak et al., 2014, 2016), as does Ana2-GFP fluorescence, indicating that both proteins turn over at the fully grown mother centrioles. In contrast, Sas-6-GFP does not detectably recover at the mother centriole, but does recover at the daughter. This suggests that once Sas-6-GFP molecules are incorporated into the cartwheel structure they do not turn over, at least not over the time course of these experiments. The underlying schematics illustrate our interpretation of the behavior of the GFP and mCherry fusions. Bars, 0.2 µm. n = 4 embryos; n = 15 centriole pairs for each protein in each experiment.

Figure 2.

Measuring the parameters of daughter centriole growth. (A) Schematic summary of the Sas-6-GFP image acquisition and processing procedure used to monitor Sas-6-GFP dynamics over time (see also Video 3). (B) Graph shows the Sas-6-GFP fluorescence intensity over time measured from four different centriole-pair tracks during nuclear cycle 12 (A.U., arbitrary units; Cent. Sep., time centriole pair separation first detected at the start of S-phase; NEB, nuclear envelope breakdown). Note that when Sas-6-GFP is imaged using spinning-disk confocal microscopy, the mother and growing daughter centrioles cannot be resolved, and so they appear as a single fluorescent focus. Therefore, to measure the growth of the daughter centriole, all centriolar Sas-6-GFP intensities were normalized by subtracting the mean initial intensity of all the mother centrioles in that embryo at the start of S-phase (so that the mean Sas-6-GFP fluorescence at the start of S-phase is 0 in every embryo). (C) Same as in B, but the mean has been taken of the fluorescence intensity of >100 centriole pairs from the same embryo. (D) The S-phase (green line) and M-phase (dark blue line) data from C were fitted by regression analysis (see Fig. S2 for a summary of the models tested). R² is used as a measure of goodness-of-fit. From this model, several parameters of centriole growth were measured for each individual embryo (as indicated in the figure). Data are represented as mean ± SD.

Figure 3.

An inverse relationship between the centriole growth rate and period sets daughter centriole size. (A) Graphs show a “mean” centriolar Sas-6-GFP incorporation profile (orange line; n ≥ 15 embryos) for embryos at nuclear cycles 11, 12, or 13. The underlying data from each embryo in this and all subsequent “mean” graphs are shown in partial opacity. The ± associated with NEB represents the SD from the mean value. A.U., arbitrary units. (B) Bar charts quantify several parameters of centriole growth derived from the mean Sas-6-GFP incorporation profiles shown in A. Statistical significance was assessed using either an ordinary one-way ANOVA test (for Gaussian-distributed data) or a Kruskal–Wallis test (****, P < 0.0001; ns, not significant). Data are presented as mean ± SD. (C) Graphs show the inverse correlation between the centriole growth rate and growth period in individual embryos during each nuclear cycle. The data were extracted from the data shown in A. Correlation strength was examined using Pearson’s correlation coefficient (r < 0.40 weak; 0.40 < r < 0.60 moderate; r > 0.6 strong), and the significance of correlation was determined by the p-value (P < 0.05). n ≥ 15 embryos for each cell cycle; n = 96, 174, and 326 centrioles (mean) per embryo in each cycle, respectively.

Figure 4.

The rate and period of daughter centriole growth are not directly determined by the length of S-phase. (A and B) Graphs show the lack of a consistent correlation between the length of S-phase and either the centriole growth rate (A) or growth period (B), in each nuclear cycle. The data were extracted from the data shown in Fig. 3 A and analyzed as described in Fig. 3 C. (C and D) Graphs and charts illustrate and quantify centriole growth parameters in WT embryos and embryos in which either the genetic dose of Cyclin B has been halved to increase the length of S-phase (Ji et al., 2004; CycB^1/2 embryos) or the genetic dose of the S-phase checkpoint protein grapes (Chk1 in vertebrates) has been halved to decrease the length of S-phase (Sibon et al., 1997; grp^1/2 embryos); ≥13 embryos were analyzed in each case. Although S-phase is ∼24% longer in CycB^1/2 embryos compared with grp^1/2 embryos, the parameters of centriole growth are essentially the same in all three types of embryos. Statistical significance was assessed using either an unpaired t test with Welch’s correction (for Gaussian-distributed data) or an unpaired Mann–Whitney test (**, P < 0.01; ****, P < 0.0001; ns, not significant). A.U., arbitrary units. n ≥ 13 embryos for each group; n = 184, 140, and 122 centrioles (mean) from WT, CycB^1/2, and grp^1/2 embryos, respectively. Data are presented as mean ± SD.

Figure 5.

Generation of the Plk4^Aa74 mutant allele. Previous studies of Plk4 have used the weak hypomorphic Plk4^c06612 allele (Bettencourt-Dias et al., 2005), so we generated a stronger allele (Plk4^Aa74) by imprecise excision of the Plk4^c06612 P-element. (A) The schematic shows the PLK4 genomic region, indicating the coding sequence (CDS; dark green), the UTRs (light green), the position of the original P{GSV1}GS3043 P-element insertion (red triangle), the position of the primers used to screen for imprecise excision of the P-element (purple), and the position of the Aa74 deletion (which deletes essentially the entire protein kinase domain; blue). (B) DNA gel shows the PCR products observed when these primers were used to amplify DNA from either WT flies or the homozygous Plk4^Aa74 deletion flies. Sequencing of the PCR products confirmed deletion of the 1,751-bp region indicated in A. MW stands for molecular weight marker as a unit label. (C) Photographs of adult flies taken without anesthesia (see also Video 4): the original Plk4^c06612 mutant flies can maintain a normal body posture and walk in a partially uncoordinated manner; Plk4^Aa74 flies are highly uncoordinated and appear to lack all proprioception, as is typical of flies lacking centrioles (Basto et al., 2006); GFP-Plk4 expression rescues the uncoordinated phenotype of Plk4^Aa74 mutant flies. (D) Micrographs show WT or Plk4^Aa74 third-instar larval brains, with DNA stained with Hoechst (blue) and centrioles revealed with GFP-PACT staining (green). Bar, 10 µm. (E) Bar chart quantifies the number of centrioles in mitotic larval brain cells of various genotypes (as indicated). n = 12 brains; n = 264 cells in toto for WT; 5 brains, 237 cells for Plk4^c06612; 7 brains, 355 cells for Plk4^Aa74; 4 brains, 143 cells for GFP-Plk4; Plk4^Aa74. Collectively, these data indicate that the Plk4^Aa74 allele behaves as a null or strong hypomorph.

Figure 6.

Plk4 levels influence the rate and period of centriole growth. (A) Graphs compare the mean Sas-6-GFP incorporation profile during nuclear cycle 12 in WT embryos (orange) and Plk4^1/2 embryos (brown). (B) Charts quantify parameters of centriole growth derived from the Sas-6-GFP incorporation profiles in A. (C–F) Graphs and charts show the same analysis for Plk4^2x (dark brown) and Plk4^RKA embryos (gray). Data are represented as mean ± SD (n ≥ 15 embryos for each group). Statistical significance was assessed using either an unpaired t test with Welch’s correction (for Gaussian-distributed data) or an unpaired Mann–Whitney test (*, P < 0.05; ***, P < 0.001; ****, P < 0.0001; ns, not significant). A.U., arbitrary units. n ≥ 15 embryos for each group; n = 92, 84, 98, and 39 centrioles (mean) from each WT, Plk4^1/2, Plk4^2x, and Plk4^RKA embryo, respectively.

Figure 7.

A second assay to measure centriole growth using the centriole distal-end binding protein GFP-Cep97. (A) Micrograph shows a 3D-SIM image of an embryo expressing Asl-GFP (a mother centriole marker), illustrating how mother centrioles in these embryos are preferentially oriented end-on to the cortex (so each mother centriole appears as a hollow ring; see also Fig. S5 and Video 5). Bar, 5 µm. (B) Micrograph shows a 3D-SIM image of a centriole pair in a fixed D. melanogaster spermatocyte expressing the distal-end-binding protein GFP-Cep97. These cells have unusually large centrioles (González et al., 1998), allowing one to easily distinguish the proximal and distal ends. The outer wall of the centriole is revealed by Asl-staining (red), which in spermatocytes is detected on both mother and daughter centrioles. GFP-Cep97 foci (green) are concentrated at the distal end of both the mother (M) and daughter (D) centriole. Bar, 0.5 µm. (C) Schematic shows how the distance between the GFP-Cep97 signal at the distal ends of the mother and daughter would be expected to increase as the centriole grows from early S-phase (d₁) to late S-phase (d₂). The schematic shows the centriole pair viewed both from the side and the top of the mother. The latter view resembles how the centrioles would usually be viewed in the early embryo, allowing us to measure daughter centriole growth in essentially 2D rather than 3D. Micrographs below show typical images of GFP-Cep97 in early and late S-phase, acquired on a Zeiss-880 Airyscan system. Bar, 0.25 µm. (D) Graph shows daughter centriole length over time, measured using this GFP-Cep97 assay in WT (n = 4 embryos; n = 69 centrioles in gray) and Plk4^1/2 embryos (n = 5 embryos; n = 84 centrioles in red). These daughter centriole growth profiles are similar to those obtained using the Sas-6-GFP incorporation assay (see Fig. 6 A). Note that centriolar GFP-Cep97 levels decrease toward the end of S-phase, so we cannot accurately track the GFP-Cep97 signal up to NEB. Data are shown as mean ± SEM; R² is used as a measure for goodness-of-fit.

Figure 8.

Sas-6-GFP is incorporated into the proximal end of the growing daughter centriole. (A) Experimental scheme to test the site of centriolar Sas-6-GFP incorporation. Embryos expressing Sas-6-GFP and Asl-mCherry were allowed to proceed through S-phase until daughter centrioles were approximately halfway through their growth period. The distance (d₁) between the centroid of the mother (calculated using the Asl-mCherry signal) and daughter (calculated using the Sas-6-GFP signal) was measured (time = T1). The centriole pair was bleached (time = T2; note that bleaching the GFP fluorescence unintentionally also bleached the Asl-mCherry fluorescence, but Asl-mCherry fluorescence at the mother rapidly recovered), and a second image acquired 1 min later (time = T3), when the distance between the centroids was measured again (d₂). The difference between d₂ and d₁ will depend upon the site of Sas-6-GFP incorporation, as illustrated. Micrographs below the schematic illustrate examples of the images acquired at the corresponding time points. Bar, 0.2 μm. (B) Graph quantifies d₁ and d₂: in unbleached centrioles, which have grown slightly between T1 and T3, d₂ is greater than d₁; after bleaching d₂ is less than d₁, indicating that Sas-6-GFP is incorporated proximally. n = 10 embryos; n = 23 centriole pairs for control (unbleached) and 24 pairs for FRAP. (C and D) These panels show similar data to that shown in A and B, but for a control experiment examining where the distal-end-binding protein GFP-Cep97 is incorporated. n = 10 embryos; n = 25 centriole pairs for control (unbleached) and 28 pairs for FRAP. In this control, GFP-Cep97 appears to be incorporated distally, as expected, indicating that these methods are sensitive enough to distinguish between the proximal incorporation of Sas-6-GFP (B) and the distal incorporation of GFP-Cep97 (D). Bar, 0.2 μm. Midlines represent the median, whiskers (error bars) mark the minimum to maximum, and bottom/top of the boxes indicate the first/third quartile of the distribution, respectively. Statistical significance was assessed using a paired t test (**, P < 0.01; ****, P < 0.0001).

Figure 9.

An analysis of Plk4 centriolar dynamics. (A) Micrographs show spinning-disk confocal fluorescence images taken from a time-lapse movie of an early embryo expressing Jupiter-mCherry (as a microtubule marker) and Plk4-GFP; time in minutes:seconds is indicated. At time t = 0:00, the embryo is in early S-phase, and the centrosomes have recently separated. Inset shows a single centriole pair (bar, 1 µm), which is then shown at 30-s or 1-min intervals (bar, 0.5 µm). The cell-cycle stage is indicated above each time interval. A series of lower-magnification views of the embryos are shown below the images of the individual centrioles to illustrate how MTs were used to determine the cell-cycle stage at each time point. Bar, 10 µm. (B) Graph shows the mean Plk4-GFP (brown) incorporation profile during S-phase of nuclear cycle 12 in WT embryos (n = 6 embryos). The mean incorporation profile of Sas-6-GFP is shown overlaid (dotted orange, same data shown in Fig. 3 A, but normalized for the length of S-phase). (C and D) Charts quantify parameters of Plk4-GFP behavior derived from mean Plk4-GFP profiles from embryos at nuclear cycles 11, 12, and 13 (n ≥ 5 embryos for each cycle; n = 34, 31, and 33 centriole pairs [mean] per embryo, respectively). A.U., arbitrary units. Data are represented as mean ± SD. Statistical significance was assessed using either an ordinary one-way ANOVA test (for Gaussian-distributed data) or a Kruskal–Wallis test, and is shown above each chart.

Figure 10.

Schematic illustration of how a Plk4-dependent homeostatic clock might set daughter centriole length in flies. (i) The schematic shows an end-on view of a mother centriole (black skeleton) in mitosis, just after it has disengaged from its daughter. Plk4 starts to be recruited to mothers by the surrounding ring of Asl (red, only shown in part) but we speculate that a small fraction of Plk4 (pink star) is stabilized by binding to Ana2 (brown)—and potentially Sas-6 (green)—so defining the site where the daughter will form. (ii) This pool of Plk4 starts to recruit more Plk4 (blue arrow, recruiting to blue ellipse); the rate of this recruitment is dependent on the amount of Plk4 initially bound to Asl and Ana2, and it sets the period of daughter centriole growth by determining how quickly Plk4 will accumulate to trigger its own destruction. (iii) When the embryo enters S-phase, the Asl/Ana2-bound fraction of Plk4 is activated (filled pink stars), allowing it to recruit more Sas-6 and Ana2 (green/brown arrow); the kinase activity of the Plk4 influences the rate of Sas-6/Ana2 recruitment, and so the rate of centriole growth. (iv) Sas-6 and Ana2 levels reach a threshold that allows cartwheel assembly, whereas the local concentration of Plk4 continues to increase. (v) Centriolar Plk4 levels reach a critical concentration that triggers its destruction—centriolar Plk4 levels start to fall, but levels of the Asl/Ana2-bound Plk4 initially remain high enough that Sas-6 and Ana2 recruitment is not slowed, and cartwheel growth continues. (vi) Eventually, levels of the Asl/Ana2-bound Plk4 are too low to support Sas-6 and Ana2 recruitment, so their concentration falls below the threshold for cartwheel growth and the daughter centriole stops growing.