Drosophila Alms1 proteins regulate centriolar cartwheel assembly by enabling Plk4-Ana2 amplification loop

Abstract

Centrioles play a central role in cell division by recruiting pericentriolar material (PCM) to form the centrosome. Alterations in centriole number or function lead to various diseases including cancer or microcephaly. Centriole duplication is a highly conserved mechanism in eukaryotes. Here, we show that the two Drosophila orthologs of the Alström syndrome protein 1 (Alms1a and Alms1b) are unexpected novel players of centriole duplication in fly. Using Ultrastructure Expansion Microscopy, we reveal that Alms1a is a PCM protein that is loaded proximally on centrioles at the onset of procentriole formation, whereas Alms1b caps the base of mature centrioles. We demonstrate that chronic loss of Alms1 proteins (with RNA null alleles) affects PCM maturation, whereas their acute loss (in RNAi KD) completely disrupts procentriole formation before Sas-6 cartwheel assembly. We establish that Alms1 proteins are required for the amplification of the Plk4-Ana2 pool at the duplication site and the subsequent Sas-6 recruitment. Thus, Alms1 proteins are novel critical but highly buffered regulators of PCM and cartwheel assembly in flies.

Keywords: Alms1; Alström Syndrome; Ana2; Centriole Duplication; Plk4.

Figure 1. Alms1a and Alms1b show different spatial-temporal localisations.

U-ExM images of (A, B) the centriolar localisation of (A) Alms1b, as shown with the expression of the fusion protein Alms1b-GFP under alms1b native promotor in alms1b deletion background (magenta; alms1b^del2, Alms1b-GFP) and (B) Alms1a, as shown with the expression of the fusion protein Alms1a-Tom under alms1a native promotor in alms1a deletion background (red; alms1a^del1, Alms1a-Tom) at early stages of spermatogenesis (spermatogonia SG; spermatocytes SC). (A) SC#1: early SC, bottom view exposing the proximal ring of Alms1b on mother centrioles (MC). SC#2: late SC, Alms1b is also detected at the proximal end of daughter centrioles (DC) (white arrow). (B) SG#1, bottom view with the focus plane on the proximal end of the MC, revealing the two concentric localisations of Alms1a. SG#2, Alms1a is also detected around the DC. Centriolar walls are revealed with acetylated α-tubulin (green, ac α-tub). MC are shown on the left, DC on the right. MC-masked insets enable better visualisation of faint localisations. (C) Alms1a-Tomato (red) localisation at centrioles in larval brain neuroblasts (NBs) and ganglion mother cells (GMCs). NB#1, shows a NB in interphase. Insets: close-up of the centriolar pair, MC at the top and DC at the bottom. Alms1a (red) forms a ring superposed to the ring of Asterless (Asl, grey) and is less abundant at the DC compared to the MC. NB#2, shows a dividing NB. The two centrioles have separated and migrated on both sides of the cell. DC in inset 1, MC in 2. (D) Alms1a-Tomato (red) localisation at centrioles in early embryos. Left panel: prophase. In the inset, the two centrioles have initiated their duplication as evidenced by the faint Alms1a concentration on the side of the centriole (white arrows) preceding the recruitment of Asl. Right panel: metaphase with advanced centriole duplication as shown in the insets by the accumulation of both Alms1a and Asl on the side of the MCs. In these two stages, the cells are dividing, forming both asters and mitotic spindle. This explains the wide and intense Asl staining with respect to Alms1a. This contrasts with NBs, GSCs and SCs where cells are not in the process of forming a spindle and where Alms1a localisation is slightly wider than that of Asl. Asl (grey) labels the centriole, α-tubulin (yellow) the cytoskeleton and Hoechst (cyan) the nuclei. In all images, mother centrioles are marked with an asterisk. In all panels, the fluorescence intensities along the white dotted line are plotted as a function of distance to further document the width of Alms1a or b localisation with respect to centriolar makers acetylated α-tubulin or Asl (A–D) Brackets show the centriolar wall width of Mother (MC) or Daughter (DC) Centrioles. Scale bars (corrected from expansion factor): (A, B): 250 nm, (C, D): 1 μm, insets: 250 nm.

Figure 2. Localisation of human ALMS1 during procentriole formation and at mature centriole.

U-ExM images of hTERT-RPE-1 cells showing (A) side and (B) bottom views of a mature centriole (green) revealing ALMS1 (detected with an antibody directed against ALMS1, magenta) localisation as a ring capping the proximal extremity of the centriolar wall. (A, B) Graphs showing the position of ALMS1 with respect to the centriolar wall in longitudinal (A) and radial (B) dimensions. Data from N = 3 independent experiments, n = 72 centrioles (A) and n = 46 (B) ALMS1 length: 107.6 nm, Tubulin length: 389.1 nm (A). ALMS1 diameter: 240.1 nm, Tubulin diameter: 241.7 nm. Error bars correspond to s.d. (B) The fluorescence intensity along the white dotted line is plotted as a function of distance. Error bars correspond to s.d. (C) Dynamics of ALMS1 recruitment at forming procentrioles. (D, E) Quantifications of ALMS1 localisation with respect to the centriolar wall revealing the onset of ALMS1 recruitment as procentrioles reach 120 nm in length. (D) Size of procentrioles without (n = 39) or with (n = 54) ALMS1 capping the proximal side. (E) Evolution of the start and end position of ALMS1 signal relative to tubulin during procentriole growth. The light green and pink regions depict the centriole length (green) and region of the centriole which is covered by ALMS1 signal (pink). Scale bars (after expansion factor correction): 100 nm. Source data are available online for this figure.

Figure 3. Alms1a is an inner PCM protein.

(A, B) U-ExM images of Alms1a-Tomato (red, in alms1a^del1;;Alms1a-Tom) relative localisation with (A) Asl (grey) and (B) Pericentrin-like protein (Plp, grey) in SC. The fluorescence intensity along the white dotted line is plotted as a function of distance. Bracket: centriolar wall width. Standard images of (C) Asl and (D) Plp in w¹¹¹⁸ (control) or alms1a^del1 in SG and SC and related fluorescence intensity quantification. In both conditions, Bld10-GFP (magenta) is expressed and used as an internal control for the quantification. Ten centrioles were quantified per stage and per testis. Here, we show two independent experiments that were quantified blind and pooled. For Asl: control: n = 70 centrioles per stage, 7 testes, alms1a^del1: n = 80 centrioles per stage, 8 testes. For Plp: control n = 80 centrioles per stage, 8 testes, alms1a^del1: n = 80 centrioles per stage, 8 testes. The box plots show the interquartile range (IQR), with the median (50th percentile) indicated by the horizontal line inside the box and the whiskers extending to the 10th and 90th percentiles. A two-sided unpaired Wilcoxon test was performed. (A, B) Scale bars (corrected from expansion factor): 250 nm; (C, D) scale bars: 1 μm. Source data are available online for this figure.

Figure 4. The RNAi depletion of Alms1 proteins leads to centriole duplication failure in various cell lineages.

(A) Unexpanded conventionally fixed testis expressing nos-Gal4>alms1^RNAi in alms1^del3 background (c) characterised by numerous centrioles per testis, a phenotype identical to the one displayed by alms1^del3 testis (a). In contrast, nos-Gal4>alms1^RNAi driven in w¹¹¹⁸ (b) shows a sharp decrease in the centriole number, with only few single centrioles observed per testis. (B) Images of conventionally fixed bam-Gal4>lacZ^RNAi or alms1^RNAi SC (outlined with white dotted lines) expressing Ana1-GFP (magenta) as a centriolar marker. Nuclei (grey). The observed distribution of the number of centrioles per spermatocytes (4C, 2C, 1C or 0 C) in the indicated genotypes (lacZ^RNAi: n = 36 cells, 3 testes; alms1^RNAi: n = 89 cells, 5 testes). bam-Gal4>lacZ^RNAi and bam-Gal4>alms1^RNAi distributions are significantly different, two-sided unpaired Fisher exact test: P = 2 × 10⁻¹⁶. (C) U-ExM images of bam-Gal4>lacZ^RNAi or bam-Gal4>alms1^RNAi centrioles in SG and SC. Centriolar walls are revealed with acetylated α-tubulin (green, tub). Asl (magenta) as a marker of PCM. On each image, numbers indicate the occurrence of the phenotype (lacZ^RNAi: n = 37 centrioles, 2 testes; alms1^RNAi: n = 34 centrioles, 2 testes). alms1^RNAi KD results in a significant loss of centriole duplication compared to lacZ^RNAi, two-sided unpaired Fisher exact test: P = 2 × 10⁻¹⁴. The fluorescence intensity along the white dotted line is plotted as a function of distance. (D) TEM images of bam-Gal4>lacZ^RNAi or bam-Gal4>alms1^RNAi mature primary spermatocytes confirming the absence of procentriole formation in alms1^RNAi. Unduplicated centrioles show no structural defects in the centriole wall (bottom right image) and induce cilia formation (bottom left and middle images). (E) Unexpanded images of wor-Gal4>lacZ^RNAi or wor-Gal4>alms1^RNAi larval brain (lacZ^RNAi: 45/45 observed NBs with centrosomes, 3 brains; alms1^RNAi: 51/55 observed NBs lack centrosomes, 5 brains). alms1 RNAi KD results in a significant loss of centriole duplication compared to lacZ^RNAi, two-sided unpaired Fisher exact test: P = 2 × 10⁻¹⁶. NBs (outlined with white dotted lines) are identified based on Miranda-GFP expression (yellow). Asl as a centriolar marker (magenta). (F) Unexpanded images of nos-Gal4>lacZ^RNAi or nos-Gal4>alms1^RNAi early embryos. β-tubulin for the mitotic spindle (magenta) and DNA (cyan). Scale bars: (A, B, E, F) 5 μm, (C) (after expansion factor correction) 250 nm, (D) 5 μm on top and middle images and 200 nm on bottom images. Source data are available online for this figure.

Figure 5. Alms1 proteins are essential for Sas-6 recruitment and required to stabilise Ana2 and Plk4.

(A) Scheme representing part of the known protein hierarchy responsible for centriole duplication, leading to procentriole formation. (B) U-ExM images of Sas-6-GFP (in magenta) in bam-Gal4>lacZ^RNAi in SG (a) before Sas-6 recruitment at the duplication site, (b) when Sas-6 is at the duplication site, (c) when procentriole has formed and in SC. (C) U-ExM images of Sas-6-GFP (in magenta) in bam-Gal4>alms1^RNAi in SG and SC. (B, C) Centriolar walls are revealed with acetylated α-tubulin (green, ac α-tub), (lacZ^RNAi: 4 testes; alms1^RNAi: 6 testes). (D) U-ExM images of Ana2-mNeonGreen knock-in (Ana2-eNG, in magenta) in bam-Gal4>lacZ^RNAi or bam-Gal4>alms1^RNAi. Centrioles at the SG stage are divided into two categories: SG short, when the length of the centrioles does not exceed their width and SG long, when their length exceeds their width (lacZ^RNAi: 3 testes; alms1^RNAi: 5 testes). The fluorescence intensity along the dotted line is plotted as a function of distance. Mother centriole (MC, on the left of all images) and daughter centriole (DC, on the right) are delineated by brackets, duplication site (ds) by an arrow. (E) U-ExM images of Plk4-meGFP knock-in (in magenta) in bam-Gal4>lacZ^RNAi or bam-Gal4>alms1^RNAi in SG (lacZ^RNAi: 3 testes; alms1^RNAi: 6 testes). Quantification of Plk4-meGFP fluorescence intensity performed on unexpended testis (lacZ^RNAi: n = 4 testes; alms1^RNAi: n = 5 testes) is shown on the right. The box plots show the interquartile range (IQR), with the median (50th percentile) indicated by the horizontal line inside the box and the whiskers extending to the 10th and 90th percentiles. Two-sided unpaired Wilcoxon test. In all images, the MC is on the left and marked with an asterisk. Scale bars (after expansion factor correction): 250 nm. Source data are available online for this figure.

Figure 6. Alms1 proteins are required for Plk4 function in centriole duplication.

(A) Images of unexpanded bam-Gal4>lacZ^RNAi or bam-Gal4>alms1^RNAi SC overexpressing Plk4 (Plk4^OE, not tagged). Asl (grey) as a centriolar marker. The different phenotypes observed are shown by genotype: 1, 2, 3 centrioles or more (> 3) per centrosome (MTOC) (lacZ^RNAi: n = 7 testes; alms1^RNAi: n = 9 testes). (B) Quantification of centrosomes with or without centriole duplication in SC overexpressing Plk4 (Plk4^OE-GFP), Alms1a (Alms1a^OE-V5) or both (Alms1a^OE-V5 +Plk4^OE-GFP) (two independent experiments are pooled, Plk4^OE-GFP: n = 734 centrioles, 5 testes; Alms1a^OE-V5: n = 505 centrioles, 5 testes; Alms1a^OE-V5 +Plk4^OE-GFP: n = 1054 centrioles, 5 testes), Fischer’s exact test, ****P = 2 × 10⁻¹⁶. (C) Images of unexpanded SC overexpressing Plk4 (Plk4^OE-GFP), Alms1a (Alms1a^OE-V5) or both (Alms1a^OE-V5 +Plk4^OE-GFP). Percentage of centrosomes with 2, 3 or more (> 3) centrioles (revealed by Asl) is indicated on the images. The overexpressed proteins are revealed with their respective tag (V5 and GFP). (D) U-ExM images of bam-Gal4 > {lacZ^RNAi, ND-Plk4-GFP} or bam-Gal4 > {alms1^RNAi, ND-Plk4-GFP}. A close-up of the region with centrosomes (white arrow on the right panel) is presented in the inset and the phenotypes observed are indicated. Acetylated α-tubulin in green as a centriolar marker. (E) Proportion of centrosomes observed with 1 to >3 centrioles per indicated genotypes (lacZ^RNAi: n = 133 centrosomes, 3 testes; alms1^RNAi: n = 55 centrosomes, 4 testes). Fischer’s exact test, P = 2 × 10⁻¹⁶. (F) U-ExM image focusing on a centrosome with the predominant composition per genotype (> 3 centrioles for lacZ^RNAi and 1 centriole for alms1^RNAi). On alms1^RNAi, the white arrow points to ND-Plk4-GFP faint concentration on the side of the remaining centriole. ND-Plk4-GFP in magenta, acetylated α-tubulin in green. (G) Images of unexpanded control or alms1^del3 SC overexpressing ND-Plk4-GFP (bam-Gal4>ND-Plk4-GFP). Overexpression of ND-Plk4-GFP in control background induces a strong overduplication of centrioles with centrosomes often containing more than 3 centrioles while alms1^del3 SC overexpressing ND-Plk4-GFP present fewer overduplicated centrioles containing less often more than 3 centrioles. Centrioles labelled with anti Asl (magenta), centrosomes with 3 or more centrioles marked with an asterisk. Scale bars: (A, C, G) 1 μm; after expansion factor correction (D) 1 μm, (D, inset, F): 250 nm. Source data are available online for this figure.

Figure 7. Alms1 proteins are required for Plk4-Ana2 interaction.

(A) Quantification of centriole number per centrosome in SC from testis overexpressing Ana2-GFP, Alms1a-V5 or both. Ana2^OE results in 13.6% of centrosomes with centriole duplication while 6.6% of centrosomes present centriole duplication upon co-overexpression of Alms1a^OE-V5 and Ana2^OE-GFP (Ana2^OE-GFP: n = 509 centrosomes, 7 testes; Alms1a^OE-V5: n = 277 centrosomes, 3 testes; Alms1a^OE-V5 +Ana2^OE-GFP: n = 652 centrosomes, 3 testes). Two-sided unpaired Fisher exact test. (B) Quantification of centriole number per centrosome in SC from testis bam-Gal4>alms1^RNAi or bam-Gal4>alms1^RNAi overexpressing ND-Plk4-GFP, Ana2-GFP or both (bam-Gal4>alms1^RNAi: n = 303 centrosomes; +ND-Plk4-GFP: n = 552 centrosomes; +Ana2-GFP: n = 753 centrosomes; +{ND-Plk4-GFP, Ana2-GFP}: n = 1348 centrosomes). Two-sided unpaired Fisher exact test. (C) Images of unexpanded SC bam-Gal4>alms1^RNAi or bam-Gal4>alms1^RNAi overexpressing ND-Plk4-GFP, Ana2-GFP or both. While only unduplicated centrioles are observed in bam-Gal4>alms1^RNAi overexpressing ND-Plk4-GFP, 10% of centrosomes contain two centrioles upon Ana2-GFP overexpression while 64% of centrosomes are composed of doublet in bam-Gal4>alms1^RNAi overexpressing both Ana2-GFP and ND-Plk4-GFP. Related controls are shown in Fig. EV6B. Asl (magenta) as a centriolar marker. Scale bar: 1 μm. (D) Model of Alms1 role in centriole duplication. We propose that Alms1 facilitates/potentializes the interaction between Plk4 and Ana2 at the onset of centriole duplication. By promoting the positive feedback loop between Ana2 and Plk4, Alms1 would hence allow for an efficient recruitment of Plk4 at the duplication site (ds), the concentration and activation of Ana2 and the subsequent formation of the cartwheel of Sas-6. In absence of Alms1, the amplification between Plk4 and Ana2 is not sufficient and Ana2 fails to recruit Sas-6. Source data are available online for this figure.

Figure EV1. Alms1a and Alms1b dynamics of localisation.

(A) Unexpended images of Alms1b-GFP (magenta) localisation prior to spermatids formation, during the second meiotic division. The two newly separated centrioles display Alms1b at their proximal end. Ana1-Tomato (green) as a centriolar marker. (B) Dynamics of localisation of endogenous Alms1a (top row, red) and endogenous Alms1b (bottom row, magenta) from spermatogonia to round spermatid stages. As observed with Alms1a-Tom transgene, endogenous Alms1a is detected from the spermatogonia until after the round spermatid stages at the proximal side of centrioles (labelled in green with Asl). As for Alms1b-GFP, endogenous Alms1b is not detected in spermatogonia. It is first observed in round spermatids in contrast to Alms1b-GFP which is detected from the late SC stage. This difference in timing could be due to technical difficulties to observe minute amounts of proteins with the antibody compared to the GFP-tagged protein. Endogenous proteins thus show dynamics of localisation at centrioles similar to that described with Alms1a-Tom and Alms1b-GFP. (C) Alms1b-GFP (magenta) localisation in neuroblasts (NB, dotted white outline) and ganglion mother cell (GMC, plain white outline). We did not detect Alms1b-GFP in NB, GMC or any other larval brain cell type. Miranda-mCherry (yellow) used as a landmark for GMC and NB. Nuclei in grey. (D) Alms1b-GFP (magenta) localisation in syncytial embryo (top) and cellularised embryo (bottom). No Alms1b-GFP is detected during the syncytial stage. Alms1b-GFP starts to localise at spindle pole during the last synchronous mitosis (cycle 14) in cellularised embryos Tubulin in green, DNA in cyan. Scale bars: (A, C, D) 5 μm, (A, inset, B) 1 μm.

Figure EV2. Asl recruitment in impaired upon loss of alms1a.

(A) Comparison of Asl intensity at centrosomes normalised to acetylated α-tubulin intensity in U-ExM control and alms1a^del1 testis. Acetylated α-tubulin labels the centriole and was used as an internal reference for Asl intensities comparison between samples. Control: n = 3 testis, SG = 20 centrosomes, early SC = 21 centrosomes; alms1a^del1: n = 3 testis, SG = 12 centrosomes, early SC = 20 centrosomes. The Box Plots show the interquartile range (IQR), with the median (50th percentile) indicated by the horizontal line inside the box and the whiskers extending to the 10th and 90th percentiles. (B) Comparison of Asl intensity at centrosomes of early SC of control, alms1a^del1 and alms1a^del1;;Alms1a-Tomato testes. Reduction of Asl intensity in alms1a^del1 is rescued to control levels upon introduction of Alms1a-Tomato transgene. Analysis based on immunofluorescence with anti Asl antibody. control: 3 testes, 15 centrosomes; alms1a^del1: 2 testes, n = 15; alms1a^del1;;Alms1a-Tomato: 3 testes, n = 15. P values were obtained with two-sided unpaired Wilcoxon test. Source data are available online for this figure.

Figure EV3. Phenotypes of alms1 mutants.

Unexpended images of SC showing centrioles in (A) alms1a^del1 (middle), alms1b^del2 (right) and in alms1b^del2 rescued by Alms1b-GFP as a control (left, similar behaviour is observed for alms1a^del1 rescued by Alms1a-Tom). Ana1-Tomato (magenta) as a centriolar marker. (B) Centrioles in control (left), alms1^del3 (middle) and alms1^del3;;Alms1a-Tom (right). For alms1^del3, a focus on engaged (a) and disengaged (b) centrioles is shown below. Asl (magenta) as a centriolar marker. (C) Quantification of centriole disjunction in control, alms1^del3 and alms1^del3;;Alms1a-Tom testes. The double alms1a,b deletion (alms1^del3) results in defective centriole cohesion for 26% of centrioles (n = 865 centrioles, 4 males). Centriole cohesion defect is fully rescued by expression of Alms1a-Tom (alms1^del3;;Alms1a-Tom) with none of the centriole pair observed presenting centriole cohesion defects (n = 666 centrioles, 3 males), as also observed in a wild-type line (100% centrioles in pairs, n = 454 centrioles, 3 males). (D) Partial rescue of the centriole duplication defect upon introduction in nos-Gal4>alms1^RNAi background of the Alms1a^R-Tom transgene. The Alms1a^R-Tom is a modified Alms1a-Tom that carries silent mutations aiming at providing resistance against alms1^RNAi mediated degradation (this study). 41% (11/27) of nos-Gal4>alms1^RNAi;;Alms1a^R-Tom testes did not show any rescue of the duplication defects. We however observed either mild or almost complete rescue of centriole duplication in respectively 48% (13/27) and 11% (3/27) of the observed testes. Bottom row: in mildly rescued testes few centrioles (marked with Asl, green) were observed in GSCs (cells contacting the hub, delineated with a white dotted line). When centrioles could be observed they were longer than in control conditions and were decorated by Alms1a^R-Tom (red) on their full length (Alms1a^R-Tom fluorescence of the boxed centriole is shown at the bottom left corner). SC (middle image) contained either no or only single centrioles and only a subset of round spermatids had a centriole (right). Top row: in fully rescued testes, centrioles co-labelled with Asl and Alms1a^R-Tom were observed in GSCs (left image, inset shows Alms1a^R-Tom localisation at the boxed centriole) and SG. SC contained pairs of centrioles (middle row) and most round spermatids contained a centriole (right). In both mildly and fully rescued testes, observed centrioles showed Alms1a^R-Tom localisation at their proximal end, as observed in control conditions. The partial rescue achieved likely reflects an incomplete resistance of the Alms1a^R-Tom construct to alms1^RNAi mediated degradation. Occurrence of rescue events by Alms1a-Tom nevertheless shows the specificity of the alms1 RNAi construct used. Centrioles labelled with anti Asl (green), Alms1a^R-Tom (red). Scale bars: (A, B) 5 μm, (B, insets) 1 μm, (D) 2 μm. Source data are available online for this figure.

Figure EV4. Alms1a localises at the proximal side of the remaining centriole in bam-Gal4>alms1^RNAi.

(A) Immunofluorescence images of bam-Gal4>lacZ^RNAi and bam-Gal4>alms1^RNAi stained with anti-Alms1a (red) and anti Asl (green) antibodies showing endogenous Alms1a localisation at the proximal end of the centrioles. (B) U-ExM images of Alms1a-Tomato (red) in bam-Gal4>lacZ^RNAi (bam-Gal4>lacZ^RNAi;;Alms1a-Tom) or bam-Gal4>alms1^RNAi centrioles (bam-Gal4> alms1^RNAi;;Alms1a-Tom). Centriolar walls are revealed with acetylated α-tubulin (green, ac α-tub). Alms1a-Tom forms a cup surrounding the proximal extremity of both mother and daughter centrioles in bam-Gal4>lacZ^RNAi as well as the proximal side of the unduplicated centriole in bam-Gal4>alms1^RNAi. Scale bars: (A) 1 μm, (B), after expansion factor correction, 250 nm.

Figure EV5. Ana2 centriolar localisation in spermatocytes.

U-ExM images of Ana2-mNeonGreen knock-in (Ana2-eNG, in magenta) in (A) bam-Gal4>lacZ^RNAi or (B) bam-Gal4>alms1^RNAi showing Ana2 localisation at centrioles in spermatocyte stage (SC). Centriolar walls are revealed with acetylated α-tubulin (green, ac α-tub), Mother centriole (MC, on the left of bam-Gal4>lacZ^RNAi images). Scale bars (after expansion factor correction): 250 nm.

Figure EV6. Overexpression of ND-Plk4-GFP in bam-Gal4>lacZ^RNAi and bam-Gal4>alms1^RNAi testes.

(A) Images of unexpanded bam-Gal4 > {lacZ^RNAi, ND-Plk4-GFP} or bam-Gal4 > {alms1^RNAi, ND-Plk4-GFP testis hub regions. In control condition, ND-Plk4-GFP (grey) forms aggregates in the cytoplasm while we only observe diffuse ND-Plk4-GFP in alms1^RNAi. The hub region is outlined with a dotted line. (B) Representative immunofluorescence images of centrosomes with or without centriole overduplication upon overexpression of ND-Plk4-GFP, Ana2-GFP or both in bam-Gal4>lacZ^RNAi genetic background. Following overexpression of ND-Plk4-GFP most centrosomes presented overduplication of centrioles including numerous events of centrosomes with more than 3 centrioles. In contrast, overexpression of Ana2-GFP induces the formation of few centrosomes with three centrioles while rosettes (> 3C) were never observed. Co-expression of ND-Plk4-GFP and Ana2-GFP leads to massive overduplication of centrioles, to an extent at least as important than observed with ND-Plk4-GFP alone. Centrioles labelled with anti Asl antibody (magenta). Scale bars: (A) 10 μm, (B) 1 μm.