The Drosophila Nol12 homologue viriato is a dMyc target that regulates nucleolar architecture and is required for dMyc-stimulated cell growth

Tissue and organ development require a precise coordination of cellular growth, proliferation, differentiation and apoptosis. At the core of the cell, and crucial for its growth, the nucleolus is the subnuclear compartment where ribosome biogenesis takes place (Boisvert et al., 2007). Cell mass accumulation is required for proliferation, implying that the regulation of nucleolar function plays an important role in the control of proliferation rates. In the nucleolus, RNA polymerase I (Pol I) synthesizes pre-rRNAs, which are processed, modified and assembled in 40S and 60S pre-ribosomal particles that are then exported to the cytoplasm. Besides its role as the ribosome factory, the nucleolus is now also considered to be a multifunctional regulatory compartment involved in RNA processing events, sensing of cell stress, and cell cycle and apoptosis regulation (Boisvert et al., 2007).

A key step in ribosome biogenesis is pre-rRNA gene transcription by Pol I, a process that in human cells is known to be stimulated by the binding of c-MYC to rRNA promoters in the nucleolus (Arabi et al., 2005; Grandori et al., 2005). Further, Drosophila Myc (dMyc; also known as Diminutive – FlyBase), which is known to control the cell cycle and apoptosis, has been shown to be necessary and sufficient for the transcription of genes encoding Pol I transcription machinery factors, such as Tif-IA and RpI135 (the largest Pol I subunit), genes encoding pre-rRNA processing and modifying factors, such as Nop60B and Fibrillarin, as well as a large set of ribosomal genes (Grewal et al., 2005; Pierce et al., 2008). The ability of dMyc to induce a coordinated nucleolar hypertrophy and to stimulate pre-rRNA transcription and ribosome biogenesis in general are required for dMyc-stimulated growth during Drosophila development (Grewal et al., 2005). Here, we identify viriato (vito) as a dMyc target gene that coordinates nucleolar and growth responses downstream of dMyc.

All crosses were raised at 25°C under standard conditions. The following stocks (described in FlyBase, unless stated otherwise) were used: w¹¹¹⁸, ptc-Gal4, hs-Gal4, Ubi-Gal4, ey-Gal4, UAS-lacZ, UAS-bsk^DN, y w hs-flp¹²²; act>y+>Gal4 UAS-GFP, Df(3L)H99, y¹ w¹¹¹⁸;; p53^[5A-1-4] (p53 null allele), rpr-11kb-lacZ, hs-GFP-Nopp140True (McCain et al., 2006), y w hs-flp¹²²;; UAS-dMyc, dm⁴/FM7(act-GFP), UAS-dicer-2, vito¹ [or pBac(RB)CG32418^e03237].

Eye-targeted RNAi knockdown of vito was induced by crossing eyeless-Gal4 with UAS-vitoRNAi [Vienna Drosophila RNAi Center (VDRC) #34548]. A second UAS-vitoRNAi transformant (named UAS-vitoRNAiKK; VDRC #102513) was also tested and observed to generate a very similar eye phenotype. Thus, the stock UAS-vitoRNAi (VDRC #34548) was used for most of the experiments.

To obtain UAS-vito transgenic lines, the full-length vito cDNA was excised from the LD10447 clone (GenBank accession AY095185) as a NotI/XhoI fragment and subcloned into NotI/XhoI-digested pUAST. Transgenic lines were generated by standard germline transformation methods and five independent lines were analysed. To obtain UAS-vito-GFP lines, the vito ORF was PCR amplified from the LD10447 cDNA using Pfu polymerase (Fermentas) and primers 5′-CACCATGACCAGGAAAAAGGCT-3′ and 5′-GTCGGCATTCCGTTTGCG-3′. The amplified fragment was cloned into pENTR/D-TOPO (Invitrogen) according to the supplier's instructions. The ORF was cloned into the pUASt-GFP vector (pTWG; a gift of T. Murphy, The Carnegie Institution of Washington, Baltimore, MD, USA) by in vitro recombination using the Gateway LR Clonase II enzyme mix (Invitrogen).

UAS-HumanNOL12-GFP flies were generated by in vitro recombination between the full ORF shuttle clone with Gateway entry vector pENTR223.1-NOL12 (clone OCABo5050F031D, ImaGenes) and the destination vector pUASt-GFP (pTWG). The pUAS-vito-GFP and pUAS-HumanNOL12-GFP constructs were verified by DNA sequencing. Transgenic lines were generated by standard germline transformation methods and four independent lines were analysed. The Ubi-Gal4 driver was used to drive low-level expression of Vito-GFP in salivary glands, and ey-Gal4 was used to drive hNOL12-GFP expression during eye development and in salivary glands. ey-Gal4 drives `leaky' Gal4 expression in the salivary glands.

Mitotic recombination was induced using the Flp/FRT method. vito knockdown clones, or clones overexpressing dMyc alone or together with vitoRNAi, were induced by heat shock (1 hour at 37°C) at 48 ± 4 hours after egg laying (AEL) and dissected at 118 ± 4 hours AEL in larvae of the genotype y w hsflp/+; act>y+>Gal4, UAS-GFP/UAS-vitoRNAi, y w hsflp/+; act>y+>Gal4, UAS-GFP/+; UAS-dMyc/+ and y w hsflp/+; act>y+>Gal4, UAS-GFP/UAS-vitoRNAi; UAS-dMyc/+.

Eye-antennal imaginal discs were prepared for immunohistochemistry using standard protocols. Primary antibodies used were: rabbit anti-cleaved Caspase-3 at 1:200 (Cell Signaling), mouse anti-Armadillo N27A1 at 1:100 [Developmental Studies Hybridoma Bank (DSHB)], mouse anti-β-galactosidase at 1:1000 (Promega, #Z3783), rat anti-Elav 7E8A10 at 1:100 (DSHB), mouse anti-Lamin ADL101 at 1:1 (DSHB), and rabbit anti-Fibrillarin at 1:250 (Abcam, #ab5821). Appropriate Alexa Fluor-conjugated secondary antibodies were from Molecular Probes. Images were obtained with the Leica SP2 confocal system and processed with Adobe Photoshop.

A digoxigenin (DIG)-labelled vito antisense RNA probe was synthesized by in vitro transcription with T7 RNA polymerase and DIG-UTP after linearisation of cDNA LD10447 with NotI and were used for whole-mount in situ hybridisation of fixed larvae. A sense RNA probe was used as a negative control. The DIG-labelled RNA probes were detected with an anti-DIG antibody coupled to alkaline phosphatase (Roche Diagnostics) using NBT/BCIP as substrate.

Female flies were transferred to 25% ethanol and incubated for 12-24 hours at room temperature. Flies were further dehydrated through an ethanol series (50%, 75%, and twice 100%; 12-24 hours each step) before being incubated twice for 30 minutes each with hexamethyldisilazane, air-dried overnight, mounted onto SEM stubs covered with carbon tape, and sputter coated with platinum. Samples were imaged using an FEI Quanta 400 microscope.

Dissected third instar salivary glands were fixed with 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer for 30 minutes and post-fixed with 4% osmium tetroxide. After washing, salivary glands were incubated with 0.5% uranyl acetate (30 minutes) and further dehydrated through a graded ethanol series (70% for 10 minutes, 90% for 10 minutes, and four changes of 100%). Salivary glands were then soaked in propylene oxide for 10 minutes and then in a mixture (1:1) of propylene oxide and Epon resin (TAAB Laboratories) for 30 minutes. This mixture was then replaced by 100% Epon resin for 24 hours. Finally, fresh Epon replaced the Epon and polymerisation took place at 60°C for 48 hours. Ultrathin sections were obtained using an ultramicrotome, collected in copper grids and then double contrasted with uranyl acetate and lead citrate. Micrographs were taken using a Zeiss EM10C electron microscope (80 kV).

For the experiment in which vitoRNAi was ubiquitously induced, RNA was isolated from control (Ubi-Gal4/+) or vitoRNAi (Ubi-Gal4/+; UAS-vitoRNAi/+) larvae 112 hours AEL. RNA was also isolated from control (w1118) and vito¹ homozygous wandering third instar larvae. Ubiquitous dMyc overexpression in wandering third instar larvae was induced by giving a heat shock (1 hour at 37°C) to hs-Gal4/UAS-dMyc larvae that were collected 4 hours after the heat shock. Control hs-Gal4/+ larvae were subjected to the same treatment.

For dMyc loss-of-function experiments, total RNA was isolated from either dm⁴ mutant or control larvae 24 hours AEL. Total RNA was also isolated from third instar salivary glands of the genotypes control (UAS-lacZ/+; ptc-Gal4/+), dMyc^OE (ptc-Gal4/+; UAS-dMyc/+), dMyc^OE+vitoRNAi (ptc>Gal4/UAS-vitoRNAi; UAS-dMyc/+) and from eye imaginal discs of genotypes control (ey-Gal4/+) and vitoRNAi (ey-Gal4/UAS-vitoRNAi). Total RNA was isolated using TRIzol (Invitrogen) according to the manufacturer's instructions and treated with DNase I (RNase-free; Ambion). cDNA was generated by reverse transcription with the SuperScript III First-Strand Synthesis SuperMix for qRT-PCR (Invitrogen). Quantitative real-time PCR analysis was performed in triplicate in 20 μl reactions containing iQ SYBR Green Supermix (BioRad), each gene-specific primer at 250 nM and 1 μl of cDNA template. Cycling conditions in a BioRad iQ5 instrument were 95°C for 3 minutes, followed by 40 cycles of denaturation at 95°C for 10 seconds and annealing for 30 seconds at 60 or 64°C depending on the primer set. Fold change relative to the expression of CaMKII, which has been used previously as a control for gene expression in dm⁴ mutants and for overexpression of dMyc in the whole larvae (Pierce et al., 2004), was calculated using the 2^–ΔΔCT method (Livak and Schmittgen, 2001). For salivary glands, data were normalised for levels of total RNA. Three to five biological replicates were analysed for each primer set. Values are presented in log₂ scale or as absolute expression levels. Statistical significance was determined using the Relative Expression Software Tool [REST; http://rest.gene-quantification.info (Pfaffl et al., 2002)]. The mathematical model is based on randomisation tests, which have the advantage of making no distributional assumptions about the data.

The following primer pairs (5′ to 3′) were used:

• CaMKII (control), TTACACCATCCCAACATAGTGC and CAAGGTCAAAAACAAGGTAGTGATAG;

• Nop60B, GAGTGGCTGACCGGTTATGT and GCTGGAGGTGCTTAACTTGC;

• p53, GCGAAAAGAAACTTCCTTAGTCTTC and TTGGGGCACGTACATATTTTAAC;

• Tif-IA, CAAGCCTATTTTCGAAGAACTTGT and CAAGGTGTCCGCTTCCAC;

• RpI135, CCCGGAGTTTAAGCAGATACC and CACATGTGGACCTCCCAAA;

• vitoqPCR1 (to detect vito transcript levels in vitoRNAi experiments), GACCAGGAAAAAGGCTCCTAA and TTGCGCTCGTTCTTAAGGTT;

• vitoqPCR2 (to detect vito transcript levels in vito¹ homozygous larvae), AGGTGAAGATCGTGGAGCTGAC and CCTGGTCGGCTTCGTCCTC;

  Fibrillarin (Fib), GCATCTCCGTTGAGACCAAT and GACACATGCGAGACTGTCGT;

• pre-rRNA (ETS), GCTCCGCGGATAATAGGAAT and ATATTTGCCTGCCACCAAAA;

• rpr, CATACCCGATCAGGCGACTC and CGGATGACATGAAGTGTACTGG;

• and puc, AGGCTATGGACGAGGATGGGTTTG and GGCGGCGAGGTCAATCTGGATG.

Eye disc areas were measured using the Straight Line tool of ImageJ 1.41o software (NIH, Bethesda, MA, USA), considering only the eye disc from the eye-antennal imaginal disc. In all the disc-proper clones assessed for clone size, nuclei were counted in the DAPI channel. Cell size estimates were obtained by dividing each clone area by the number of nuclei in the clone. For the comparison of nuclear, cytoplasmic and cell sizes in different genetic backgrounds, the radius (r) was measured using the Straight Line tool of ImageJ. Volume (V) was approximated as V=4/3 π r³. For each genotype, 62-82 nuclei were scored. GraphPad Prism 5.0 was used for statistical analysis and for generating the graphical output. Statistical significance was determined using an unpaired two-tailed Student's t-test, with a 95% confidence interval, after assessing the normality distribution of the data with the D'Agostino-Pearson normality test.

In a search to identify genes required for tissue growth in Drosophila, we observed that expression of an RNAi transgene targeting the uncharacterised CG32418 gene [which we named viriato (vito)] in the developing eye resulted in a reduced and rough eye (Fig. 1A and see Fig. S1 in the supplementary material). When we examined the effects of vitoRNAi expression in the eye primordium, we observed that although posterior retinal differentiation still occurred, a very significant reduction in eye disc size was visible, particularly in the anterior proliferative region of the disc (Fig. 1B and see Fig. S1 in the supplementary material). This suggested that vito is required for growth and/or proliferation before the onset of photoreceptor differentiation. In support of this hypothesis, although widely transcribed in imaginal discs, vito expression is stronger in the anterior region of the eye disc (Fig. 1C and see Fig. S2 in the supplementary material). Similar phenotypes in eye discs and adults were observed with a second RNAi line (UAS-vitoRNAiKK) and in eye discs homozygous for the vito¹ allele (vito¹ results from a PiggyBac insertion, PBac{RB}CG32418^e03237) (Fig. 1 and see Fig. S1 in the supplementary material). In addition, reduction of vito levels, either in vito¹ homozygous larvae or induced by generalised expression of vitoRNAi, caused a significant developmental delay (see Fig. S3 in the supplementary material). Very significant reductions of vito mRNA levels were detected by qPCR in larvae expressing vitoRNAi and in vito¹ homozygous larvae (Fig. 1E). To confirm the specificity of the phenotypes observed, we attempted to rescue the effects of vitoRNAi by co-expressing a full-length vito cDNA. Indeed, eye disc size (Fig. 1F and see Fig. S1 in the supplementary material), adult eye size and morphology and several other cellular parameters (see below, and see Fig. S1 in the supplementary material) were significantly rescued by supplying vito back.

Sequence comparison analysis predicts vito to encode a small protein with an N-terminal Nol12 domain that is conserved in the Nol12 protein family (Fig. 1G). The Drosophila genome contains only one non-redundant gene for Nol12, as do all the genomes found to encode Nol12 proteins. Therefore, we attempted to rescue the reduced growth induced by vitoRNAi by co-expressing the single human NOL12 (hNOL12) protein. Remarkably, hNOL12 was also able to fully rescue eye disc growth, suggesting that there is an evolutionary conservation of Nol12 protein family functions in tissue growth (Fig. 1F and see Fig. S1 in the supplementary material). We concluded from these results that inducible vitoRNAi was efficient and specific, which, together with the ease and flexibility of vito knockdown in both clones and targeted tissues, led us to mainly use vitoRNAi for the experiments described here.

Human NOL12 has been identified in several proteomic analyses of human nucleoli (Ahmad et al., 2008). Mouse NOL12 localises to nucleoli in COS7 cells, and knocking down NOL12 function in this cell line results in nucleolar fragmentation (Suzuki et al., 2007). In budding yeast, the Nol12 homologue Rrp17p/Ydr412p has been implicated in nucleolar non-coding RNA processing events (Peng et al., 2003; Li et al., 2009; Oeffinger et al., 2009). Owing to the relatively low conservation of the Nol12 domain, we analysed whether Vito localised to the nucleolus. Targeted expression of low levels of GFP-tagged Vito in Drosophila tissues showed colocalisation of Vito with Fibrillarin (Fig. 1H; and also in S2 cells, see Fig. S4 in the supplementary material), a nucleolar methyltransferase required for pre-rRNA modification that localises to the fibrillar region of Drosophila nucleoli (Orihara-Ono et al., 2005). Consistent with its ability to functionally compensate for vito knockdown, hNOL12 also localised to the nucleolus when expressed in Drosophila larval tissues (Fig. 1H).

Since no in vivo functional information is thus far available for members of the Nol12 family in any metazoan, we examined the role of vito in tissue growth during Drosophila development. Using Flp/FRT-based recombination, we induced during early larval development clones of cells expressing vitoRNAi that were analysed later in third instar eye imaginal discs. During this period, most cells in the disc are actively proliferating. In the eye disc, vitoRNAi-expressing clones were significantly smaller than control neutral clones, both in the peripodial epithelium and in the eye disc proper (Fig. 2A-C). This was not due to a reduction in cell size (Fig. 2D and see Fig. S5 in the supplementary material), suggesting that in proliferating cells vito knockdown slows down proliferation while maintaining the coordination of cell growth and cell cycle.

Next, we investigated whether vito misregulation could also result in apoptosis, contributing to the observed defect in tissue growth. In contrast to wild-type eye discs, where apoptosis is virtually absent during larval development, vitoRNAi eye discs exhibited a significant number of cells undergoing apoptosis, as detected by the presence of activated cleaved Caspase-3 (Decay – FlyBase) (Fig. 3A,B), and this was accompanied by a substantial reduction in disc size (42%, P<1×10^–4; Fig. 3C). Therefore, vito is required for cell survival.

To determine whether the reduction in disc size could be explained exclusively by this increase in apoptosis, vitoRNAi was induced in a heterozygous background for the H99 deficiency (Fig. 3A), which deletes the three crucial proapoptotic genes reaper (rpr), grim and hid (Wrinkled – FlyBase) (White et al., 1994; Grether et al., 1995; Chen et al., 1996). In this genotype, in which apoptosis was almost completely suppressed (Fig. 3B), the average size of the eye disc was only partly rescued compared with vitoRNAi discs (22%, P=0.0072), but was still significantly smaller than in control discs (29% reduction, P<1×10^–4) (Fig. 3C). Similar results were obtained when vitoRNAi was co-induced with the baculovirus caspase inhibitor p35 (see Fig. S6 in the supplementary material). Furthermore, even though we only detected a slight, non-statistically significant increase in rpr transcript levels by qPCR in whole eye-antennal vitoRNAi discs (Fig. 3D), we observed a robust and localised upregulation of the rpr-11kb-lacZ transcriptional reporter (Nordstrom et al., 1996) in the anterior proliferative domain of the eye disc (Fig. 3E). Interestingly, most of the apoptotic death was detected in the anterior domain, where we had also detected the strongest vito transcription. The fact that only a fraction of cells within the eye disc upregulated rpr (as monitored by the rpr-11kb-lacZ reporter) might have masked the increase in rpr transcripts in whole discs measured by qPCR.

Overall, these results indicate that the growth deficit induced by reducing vito function cannot be explained by a generalised induction of apoptosis. Since Vito is a nucleolar protein and nucleolar stress in mammalian cells leads to apoptosis mediated by p53 stabilisation (Rubbi and Milner, 2003; Yuan et al., 2005), we tested whether apoptosis in vitoRNAi eye cells was dependent on p53 function. Expression of vitoRNAi still caused a significant level of apoptosis in a p53 null mutant background (Fig. 3A,B), or when co-expressed together with a dominant-negative form of p53 (R155H) (see Fig. S6 in the supplementary material) that is able to block p53 pro-apoptotic activity (Ollmann et al., 2000). The absence of p53 function did not rescue normal tissue growth in vitoRNAi eye discs (Fig. 3C). In addition, we could not detect elevated p53 transcript or protein levels in vitoRNAi eye discs (Fig. 3D and see Fig. S6 in the supplementary material).

In Drosophila, as well as in mammals, the c-Jun N-terminal kinase (JNK) pathway has been implicated in apoptotic cell death in a variety of cellular and developmental contexts (reviewed by Igaki, 2009), and has been suggested to act both upstream (McEwen and Peifer, 2005) and downstream (Kuranaga et al., 2002) of rpr function. We studied the possible involvement of JNK signalling in apoptosis resulting from vitoRNAi expression by: (1) co-expressing a dominant-negative form of JNK (BskDN) (Fig. 3A); and (2) analysing puc transcript levels (Fig. 3D) and puc-lacZ reporter transcription (see Fig. S6 in the supplementary material) as a read-out for JNK signalling (Martin-Blanco et al., 1998). The results of these experiments suggest that upon vito knockdown there is a weak activation of JNK signalling, as puc mRNA levels in eye discs increased by 20% (but in a non-significant manner, P=0.059) (Fig. 3D), and the co-expression of BskDN resulted in minor apoptotic rescue (Fig. 3A,B). We also failed to detect any upregulation of puc-lacZ in vitoRNAi clones in eye discs (see Fig. S6 in the supplementary material), suggesting that activation of JNK signalling is not the major pathway that mediates the apoptotic process resulting from vito knockdown.

Overall, these experiments show that vito is required for tissue growth during Drosophila development. This requirement can be explained only in part by the role of vito in preventing caspase-mediated apoptosis, in which the JNK pathway itself might play a marginal role.

To investigate further the mechanisms by which Vito, a nucleolar protein, might be controlling tissue growth, we analysed the effects of reducing or increasing Vito levels on the nucleolus of salivary gland cells, which, owing to their polyploidy, have large nucleoli. As in the eye disc, the reduction of vito expression in the salivary glands also caused a substantial deficit in tissue growth (Fig. 4A,B). To examine nucleolar structure and activity we looked at the distribution of Fibrillarin and Drosophila Nopp140-True (Cui and DiMario, 2007). The human NOPP140 (NOLC1) protein is a conserved phosphoprotein that interacts with the RNA Pol I 194 kDa subunit (RPA194) and is proposed to play a role in the maintenance of nucleolar structure (Chen et al., 1999). NOPP140 is also required for the assembly or recruitment of small nucleolar ribonucleoproteins (snoRNPs) to the nucleolus, where these complexes guide site-specific 2′-O-methylation and pseudo-uridylation of pre-rRNA (Wang et al., 2002). In Drosophila, as previously described (McCain et al., 2006), a GFP-tagged Nopp140-True protein expressed from a heat shock-inducible promoter was detected uniformly in the nucleolus of salivary gland cells, where it colocalised with Fibrillarin (Fig. 4D). In vitoRNAi-expressing cells, Nopp140-True still localised to the nucleolus but its pattern became more dispersed and `hollow', concentrating at the nucleolar periphery. This redistribution paralleled the redistribution of Fibrillarin, which was concentrated in the nucleolus at significantly higher levels (Fig. 4E; for similar results with vito¹ mutants and rescue with Vito and hNOL12, see Fig. S7 in the supplementary material). In vito-overexpressing cells (Fig. 4C), Nopp140-True was still recruited to the nucleolus, which became rounded and displayed large vacuoles devoid of Nopp140 signal. Interestingly, Fibrillarin was absent from Vito-overexpressing nucleoli and accumulated at low levels in the nucleoplasm (Fig. 4F, and see Fig. S4 in the supplementary material for similar results in S2 cells).

These changes in nucleolar morphology resulting from altering Vito expression levels were confirmed by TEM (Fig. 4G-I). As previously noted (Orihara-Ono et al., 2005), the wild-type Drosophila nucleolus does not display the characteristic vertebrate tripartite organisation, as only a homogeneous region with a regular surface is observed (Fig. 4G). Knocking down vito resulted in a substantial reduction in the packaging of nucleolar components, and the nucleoli exhibited a clear overall granular organisation (Fig. 4H). By contrast, Vito overexpression induced a dramatic re-organisation of the nucleolus into two distinct regions: a peripheral and highly compact region with very smooth borders that surrounded internal vacuole-like structures (Fig. 4I).

These results show that Vito is a major regulator of nucleolar architecture and that it regulates the recruitment of specific nucleolar components: whereas Nopp140 remains associated to the nucleolus when Vito levels are changed, the recruitment of the snoRNP methyltransferase Fibrillarin depends critically on Vito. Interestingly, in human cells, NOPP140, but not fibrillarin or the rRNA pseudo-uridylase dyskerin, can interact with ribosomal gene chromatin independently of ongoing Pol I transcription (Prieto and McStay, 2007). This suggests that Vito could act to control the recruitment or retention of a second layer of proteins, such as Fibrillarin, that require binding to nucleolar hub proteins (Emmott and Hiscox, 2009) in order to associate with the nucleolar compartment.

We next aimed to establish correlations between the observed nucleolar alterations and tissue growth deficit caused by vito misregulation in salivary gland cells. Firstly, quantification of estimated salivary gland cell volumes revealed that both vitoRNAi and Vito-overexpressing gland cells were significantly smaller than controls (51% and 77% reductions, respectively; P<1×10^–4; Fig. 4J). However, we noted significant differences in the impact of increasing or decreasing vito on cytoplasmic and nuclear sizes. The size reduction in vitoRNAi-expressing cells was basically due to a reduction in cytoplasmic volume (55%, P<1×10^–4), as nuclear volumes were not significantly reduced (20%, P=0.0228). This suggests that vito knockdown mainly affects cytoplasmic mass accumulation, while not interfering with the endoreplication process, something that we also assessed by quantifying DAPI staining intensity (see Fig. S7 in the supplementary material). Vito overexpression caused not only a substantial reduction in cytoplasmic volume (81% P<1×10^–4), but also a decrease in nuclear size (48%, P<1×10^–4). However, this nuclear size reduction might be secondary to the dramatic loss of cytoplasmic mass: it is likely that the extremely aberrant nucleolar morphology compromises basic cellular processes such as protein synthesis, indirectly affecting DNA endoreplication (and hence nuclear size), as these two processes are intimately linked (reviewed by Edgar and Orr-Weaver, 2001).

Our results showing that vito is required for the proliferation of diploid imaginal cells and for the growth of polyploid salivary gland cells led us to hypothesize that vito could act downstream of dMyc, a crucial regulator of Drosophila growth (Johnston et al., 1999) that is also known to be a major regulator of nucleolar growth (Grewal et al., 2005). Since dMyc stimulates tissue growth in part by activating the transcription of genes required for nucleolar function and ribosome biogenesis (Grewal et al., 2005; Demontis and Perrimon, 2009), we tested whether vito transcription was under dMyc control. Indeed, dMyc strongly regulated vito mRNA levels (Fig. 5). When quantified by qPCR, vito transcript levels were reduced by 85% in larvae homozygous for the null dMyc^dm4 allele, and increased 1.9-fold and 18.9-fold upon dMyc overexpression in whole larvae and in salivary glands, respectively (Fig. 5A).

We next assessed whether vitoRNAi affected the transcriptional response of several known dMyc targets (Grewal et al., 2005; Pierce et al., 2008). The set of targets analysed included genes encoding factors involved in pre-rRNA transcription (the Pol I subunit RpI135 and the basal Pol I Tif-IA transcription factor), genes encoding pre-rRNA processing/modifying enzymes [the pseudo-uridylase Nop60B (the Drosophila homologue of dyskerin) and Fibrillarin] and also primary pre-rRNA transcripts [containing the external transcribed spacer (ETS) region]. As expected, dMyc overexpression induced strong upregulation of all these target transcripts in salivary glands. However, co-expression of vitoRNAi did not affect their strong induction by dMyc (Fig. 5B). Therefore, vito does not appear to be required for the transcriptional induction of dMyc target genes involved in nucleolar growth and ribosome biogenesis.

To test whether vito function is required for dMyc-stimulated growth, we overexpressed dMyc in the diploid proliferative eye disc cells, in the presence and absence of vitoRNAi. Whereas mitotic clones expressing dMyc were significantly larger than control clones, the co-expression of dMyc with vitoRNAi resulted in a reduction in clone size relative to control levels (Fig. 5C,D). This reduction was the consequence of a partial reversion of the dMyc-induced cell size increase together with a decrease in the number of cells per clone (Fig. 5E,F). Similarly, in polyploid salivary gland cells, dMyc overexpression on its own dramatically increased nuclear and overall cell size (Pierce et al., 2004) (Fig. 6A,B). However, when vitoRNAi was co-expressed cell growth was considerably attenuated (Fig. 6A,B). This resulted from significant reductions in cytoplasmic (down by 59%, P<1×10^–4) and nuclear (down by 70%, P<1×10^–4) volumes when compared with dMyc overexpression (Fig. 6B). Therefore, genetically, vito lies downstream of dMyc in the control of cell growth and proliferation.

To identify the cellular function that vito performs during dMyc-stimulated cell growth, we examined whether Vito regulated nucleolar organisation when dMyc was overexpressed. As previously described (Grewal et al., 2005), dMyc overexpression in salivary gland cells led to an enlarged nucleus and nucleolus (Fig. 6A). When compared with the wild type, these nucleoli displayed slightly less compact packaging, with closely intermingled fibrillar and granular components. Crucially, when dMyc overexpression was induced together with vitoRNAi, the nucleolus consistently displayed a segregated organisation, with a peripheral granular region surrounding an internal fibrillar area (Fig. 6A). This nucleolar structural arrangement resembles, in part, the structure of the vitoRNAi-only nucleolus, although in the latter case the nucleolus shows a more decondensed granular region (compare Fig. 6A with Fig. 4H). Although these experiments involve partial knockdown and not null mutations, precluding definitive statements about epistasis, overall our findings suggest that vito lies downstream of dMyc in the control of Drosophila nucleolar structure.

Our data identify the Nol12 homologue vito as an important regulator of dMyc function in the stimulation of nucleolar biogenesis and mass accumulation during Drosophila development. We show that dMyc controls vito mRNA levels to regulate nucleolar architecture and that vito is required for dMyc to reach its full potential as a potent cell growth inducer. Furthermore, the knockdown of vito expression also correlated with an increase in p53-independent, caspase-mediated apoptotic cell death, suggesting a potential novel link between structural and functional changes in the nucleolus and activation of the pro-apoptotic rpr/grim/hid complex.

During development, dMyc plays a crucial role in translating intracellular and extracellular cues to regulate the pace of cell growth and proliferation. One of the main mechanisms for dMyc-stimulated growth appears to be the transcriptional control of nucleolar ribosome biogenesis genes (Grewal et al., 2005; Hulf et al., 2005; Teleman et al., 2008; Demontis and Perrimon, 2009). Cells of the salivary glands are polyploid secretory cells with very active biosynthetic pathways. In these cells, increasing or reducing Vito levels results in changes in the nucleolar localisation patterns of the pre-rRNA methyltransferase Fibrillarin and in alterations in nucleolar structure. The fact that vito does not appear necessary for the expression of dMyc targets implicated in ribosomal biogenesis suggests that part of the control that dMyc exerts on the nucleolus is mediated independently of its regulation of vito expression. In addition, several results support the hypothesis that the Myc-Nol12 regulatory relationship is evolutionarily conserved. Genome-wide chromatin immunoprecipitation analysis has shown that c-MYC binds the NOL12 promoter in both a human transformed B-cell line (Zeller et al., 2006) and in mouse stem cells (Kim et al., 2008). We have also identified non-canonical E-box motifs (CACATG) (Zeller et al., 2006) in the putative proximal promoter regions of both vito and human NOL12 (data not shown).

In addition to its function in cell growth downstream of dMyc, vito plays a role in supporting the proliferation and survival of diploid cells. dMyc mutants are smaller than the wild type, and dMyc mutant cells grow poorly in the context of wild-type tissue (Johnston et al., 1999; Moreno and Basler, 2004; Benassayag et al., 2005; Wu and Johnston, 2010). Therefore, vito is a rate-limiting factor for tissue growth that links dMyc with nucleolar architecture. The mechanisms enacting this link might prove relevant for the regulation of Myc function in tumourigenesis (Meyer and Penn, 2008).

We thank Corinne Benassayag, Filipe Josué, Hermann Steller, Patrick DiMario, Robert Eisenman, the Bloomington Drosophila Stock Center, the Vienna Drosophila RNAi Center, and the Developmental Studies Hybridoma Bank for reagents; Inês Barbosa and Sofia Pinho for technical assistance; and Hélder Maiato and Vitor Barbosa for critical reading of the manuscript. This work was supported by grants PPCDT/BIA-BCM/56043/2004 (FCT, Portugal), and BFU2006-00349, BFU2009-07044 and Consolider `From Genes to Shape' (MICINN, Spain) to F.C. J.M. was funded by FCT (SFRH/BD/33182/2007) and P.S.P. was funded by `Programa Ciência' (FCT).