Developmental roles of Drosophila tRNA processing endonuclease RNase Z^L as revealed with a conditional rescue system

Abstract

Drosophila RNase Z^L (dRNaseZ) belongs to a family of endoribonucleases with a major role in tRNA 3′-end processing. The biochemical function of RNase Z^L is conserved from yeast to human. Here we present a study of its biological function during Drosophila development. In flies, dRNaseZ provides a non-redundant function, as the RNZ^ED24 knockout (KO) mutation causes early larval lethality. Mosaic and conditional rescue techniques were employed to determine dRNaseZ requirements at later stages. We found that dRNaseZ activity is essential for all phases of fly development that involve cell division, including growth of adult tissue progenitors during larval and metamorphic stages, and gametogenesis in adults. At the cellular level, two major phenotypes were identified – cell growth deficiency in endoreplicating tissues and cell cycle arrest in mitotic tissues. While cell growth and proliferation are both dependant on protein synthesis, the two phenotypes displayed reliance on different dRNaseZ functions. We found that dRNaseZ KO completely blocks tRNA maturation without diminishing the abundance of mature tRNA molecules. Our data indicate that growth arrest of endoreplicating cells is primarily attributed to the relocation of the pool of mature tRNAs into the nuclei causing a decrease in translation efficiency. Mitotically dividing cells appear to be less dependent on translation machinery as they maintain their normal size when deprived of dRNaseZ activity, but rather display a cell cycle arrest at the G₂-M transition.

INTRODUCTION

The RNase Z enzyme is a highly conserved single-strand endoribonuclease expressed in all living cells. There are two classes of RNase Z proteins, long RNase Z^L and short RNase Z^S (Vogel et al., 2005). The RNase Z^S form has been found in prokaryotes, eukaryotes and archaea, while RNase Z^L – only in eukaryotes. All RNase Z proteins belong to the metallo-β-lactamase (MBL) superfamily, as they require divalent metal ions and feature an MBL fold forming a catalytic core of the enzyme. Crystal structure analysis of three different bacterial RNase Z^S (Li de la Sierra-Gallay et al., 2005; Ishii R et al., 2005; Kostelecky et al., 2006) and homology modeling of human RNase Z^L (Redko et al., 2007) showed that the short form of enzyme functions as a homodimer, and the long one – as a monomer.

RNase Z catalyzes hydrolysis of a phosphodiester bond producing 3′-hydroxy and 5′-phospho termini as it participates in tRNA maturation by cleaving off a 3′ trailer sequence (Mayer et al., 2000). The tRNA 3′ processing activity of RNase Z was used to trace the protein when it was originally purified from wheat germ to homogeneity, which then led to the cloning of the very first RNase Z gene from Arabidopsis thaliana (Schiffer and Marchfelder, 2002). Based on sequence homology, many more RNase Z enzymes were identified and cloned from other organisms. All of them display tRNA 3′ processing activity in vitro (Schiffer and Marchfelder, 2002; Takaku et al., 2003; Spath et al., 2005) and some of them were shown to participate in the tRNA maturation pathway in vivo (Pellegrini et al., 2003; Dubrovsky et al., 2004; Zhao et al., 2009; Xie et al., 2011). Importantly, identification and cloning of RNase Z has allowed a detailed biochemical characterization of the enzyme revealing that RNase Z could cleave a broader spectrum of substrates including coding and noncoding RNA. The Saccharomyces cerevisiae RNase Z^L appears to function in biogenesis of rRNA (Peng et al., 2003; Chen et al., 2005). Human RNase Z^L cleaves MALAT1 (metastasis-associated lung adenocarcinoma transcript 1) at the 3′ end to yield a 61-nt noncoding mascRNA (MALAT1-associated small cytoplasmic RNA), whose function remains unknown (Wilusz et al., 2008). In plants, RNase Z^L is required for the synthesis of several small nucleolar RNAs (snoRNAs). While most of snoRNA encoding genes are polycistronic, there are twelve tRNA-snoRNA dicistronic genes in A. thaliana. They are transcribed by RNA Polymerase III (Pol III) generating a primary transcript that is processed by RNase Z^L to liberate both the tRNA and snoRNA (Kruszka et al., 2003; Barbezier et al., 2009). Mammalian RNase Z^L mediates biogenesis of certain viral microRNAs (miRNAs). Nine murine γ-herpesvirus 68 (MHV68) miRNAs are located downstream of tRNA-like sequences, and each is transcribed from the Pol III promoter as one tRNA-miRNA fusion transcript (Pfeffer et al., 2005). These primary transcripts are cleaved by RNase Z^L resulting in nonfunctional tRNA-like molecules and viral pre-miRNA hairpins ready for nuclear export and Dicer processing (Bogerd et al., 2010). Finally, it has been shown that in vitro RNase Z^L can cleave any mRNA target under the direction of a small guide RNA (sgRNA) that would fold into a structure resembling tRNA fragments (Nakashima et al., 2007). A group of sgRNAs that could complex with RNase Z^L and program its endonucleolytic activity in vivo includes 5′-half or 3′-half tRNA fragments, rRNA and snRNA fragments. For example, when charged with the 5′-half-tRNA^Glu, human RNase Z^L downregulates expression of the PPM1F gene, when it is charged with the 28S rRNA fragment, it downregulates DYNC1H1 gene expression (Elbarbary et al., 2009). Thus, results from a number of studies clearly establish the ability of RNase Z^L to process substrates other then tRNAs, and suggest participation of the enzyme in a wider array of biological pathways than previously anticipated.

There have been only few studies attempting to assess functional relevance of RNase Z^L in live organisms. The fission yeast Schizosaccharomyces pombe has two RNase Z^L genes. Both of them appeared to be vital as they encode proteins that are differentially targeted to the nucleus and mitochondria (Zhao et al., 2009; Gan et al., 2011). In Saccharomyces cerevisiae, the lone Trz gene encoding RNase Z^L is also essential (Tavtigian et al., 2001). It is noteworthy that although deletion of the yeast RNase Z^L genes is lethal, their tRNA 3′-end processing function is dispensable as both S. pombe and S. cerevisiae have a backup exonucleolytic pathway for tRNA maturation (Maraia et al., 2011). What makes yeast RNase Z^L vital is not known. Genome of A. thaliana harbors four genes – two for RNase Z^L and two for RNase Z^S. All four RNase Z isoforms exhibit tRNA 3′ processing activity when tested in vitro, but only one of them is essential in vivo as its deletion causes embryonic lethality (Canino et al., 2009). Given that four RNase Z proteins of A. thaliana display different distribution among subcellular organelles, it was suggested that the essential isoform, which is the only one located in chloroplasts, has a unique function that cannot be replaced by other nucleases. In Caenorhabditis elegans, RNase Z^L appears as a nonvital gene that plays role in gametogenesis (Smith and Levitan, 2004). Knockdown of hoe-1, the C. elegans homolog of RNase Z^L, delays worm development and produces sterile adults resulting from a cell cycle arrest in the germline. While these studies suggest that RNase Z^L is implicated in basic cellular processes such as growth and proliferation, our knowledge on biological functions of RNase Z^L is still limited. It remains to be determined how RNase Z^L contributes to the in vivo development of a multicellular eukaryotic organism.

Previously, we reported the identification and biochemical analysis of dRNaseZ in flies, the lone Drosophila homolog of RNase Z^L (Dubrovsky et al., 2004). Knockdown of dRNaseZ by RNAi impaired larval growth and development causing death after the second-to-third instar molt (Xie et al., 2011). To clarify further the role of dRNaseZ in fly development, we have now isolated and characterized the knockout allele. In this report, we have utilized a conditional rescue system to uncover distinct requirements for dRNaseZ in postmitotic, mitotic and endoreplicating tissues at different stages of fly development.

MATERIALS AND METHODS

Fly stocks

Crosses were carried out at 25 °C in vials containing standard cornmeal-molasses-agar medium unless otherwise stated. The following flies from the Drosophila Stock Center (Bloomington, IN, USA) were used in this study: P{SUPor-P}KG07814 (FBst0014659), Sp/CyO;Sb,P{Δ2–3}99B/TM6B (FBst0003629), Df(2R)12/CyO (FBst0005425), FRT^G13,L/SM6a (FBst0001958), FRT^G13,GFP^nls (FBst0005826), FRT^G13,ovo^D1/S,Sp,Ms(2)M,bw^D/CyO (FBst0002125), hsFLP¹²;Sco/CyO (FBst0001929), L,Pin/CyO,GFP (FBst0005194), tubGal4/TM3,Sb (FBst0005138), and Kr/CyO;D/TM6C,Sb,Tb (FBst0007199). All marker mutations, balancer chromosomes, and cytological positions are described in the FlyBase (www.flybase.org).

P element mobilization

Deletions of the dRNaseZ gene were generated from the P{SUPor-P} line using method of imprecise excision mutagenesis (Engels et al., 1990). Based on the sequencing data released by the Gene Disruption Project, the KG07814 transposon is located 21 bp upstream from the dRNaseZ transcription start site. To initiate excision, P{SUPor-P} homozygotes were crossed to flies of the P{Δ2–3}99B stock that carry a stable transposase source. F0 jump-starter males w/Y;P{SUPor-P}/CyO;P{Δ2–3}99B/+ were collected and crossed individually to females w;Kr/CyO. Flies carrying an excision were identified as white-eyed F1 males. To recover imprecise excisions removing flanking dRNaseZ gene, F1 males were screened for hemizygous lethality in F2 with the Df12 deletion that uncovers the chromosomal interval 46F1-47A10 encompassing the dRNaseZ locus.

Lethal stage determination

The dRNaseZ knockout alleles were maintained in stocks with the CyO,GFP balancer chromosome. Flies were mated in population cages and allowed to lay eggs on apple-juice/agar/yeast plates for 2 h. After further 18–20 h of embryonic development at 25 °C, synchronously hatching control (RNZ^ED24/CyO,GFP) and mutant (RNZ^ED24/RNZ^ED24) larvae were hand-collected using GFP as a selection marker, scored, and placed in groups of 50 animals on standard cornmeal media in 35-mm Petri dishes. The number of larvae was scored daily, using mouth hook morphology as a developmental stage marker. At least 300 larvae were studied per genotype.

Transgenic rescue

The UAS-RNZ-V5 rescue-construct and corresponding transgenic flies were described previously (Xie et al., 2011). To generate the hs-RNZ-V5 (hs-RNZ) construct, the dRNaseZ coding sequence was joined at the C-terminal end with the V5-epitope sequence and placed under the heat-inducible promoter in the pCaSpeR-hs vector. Transgenic flies were generated by P element-mediated germline transformation. Several homozygous viable transgenic lines carrying this construct on 3^rd chromosome were established. Transgene expression was activated with a 1-hour heat-shock (HS) treatment at 37 °C; full-size protein synthesis was confirmed by Western blot hybridization with the anti-V5 antibody (Invitrogen).

To generate a genomic rescue-construct, a 6.6-kb BglII fragment containing dRNaseZ coding DNA along with the 3-kb upstream and 1-kb downstream sequences was amplified on the BAC clone (RP98-48F07 from BPRC, CHORI) template with primers 5′-A AGA TCT AAC ATC ACT AAT CCA GAG-3′ and 5′-T AAC ATG AGA TCT TCT TTA GGT ACC-3′ and cloned with the pGEM-T Easy vector system (Promega). With the Site-Directed Mutagenesis Kit (Stratagene) and two primers (5′-CGA AAA CTG GCA GAA ACC GGT TGA TTG CCG G-3′ and 5′-C CGG CAA TCA ACC GGT TTC TGC CAG TTT TCG-3′) we created a unique AgeI restriction site right in front of the dRNaseZ stop codon, where we inserted a V5-tag coding sequence. After confirming the sequence integrity of the whole genomic fragment, it was cut out with BglII, purified and ligated into the BamHI site of the pCa4B2G vector (a gift from Dr. Perrimon at Harvard, Boston, MA) yielding the genRNZ-V5 rescue-construct. To establish a transgenic line, we used phiC31-mediated site-specific integration to insert the transgene into the attP site on 3^rd chromosome at 68A (Groth et al., 2004). Microinjection of embryos with DNA constructs was performed by Model System Genomics (Duke University, Durham, NC).

For rescue experiments with the UAS-RNZ-V5 construct, RNZ^ED24/CyO;UAS-RNZ flies were crossed to RNZ^ED24/CyO;tubGal4/TM3,Sb. For rescue with genRNZ-V5, RNZ^ED24/CyO flies were crossed to RNZ^ED24/CyO;genRNZ/TM3,Sb. For rescue with hs-RNZ-V5, offspring of the RNZ^ED24/CyO flies crossed to RNZ^ED24/CyO;hs-RNZ/TM3,Sb were subjected to 1-h HS every 24 h from day one after egg deposition (AED) till adult eclosion. For each construct the rescue efficiency was calculated by comparing the number of the rescued progeny to their heterozygous siblings.

In situ hybridization

Whole-mount in situ hybridization was performed basically according to Tautz and Pfeifle (1989) following previously described protocol (Dubrovsky et al., 2002). Antisense and sense (control) RNA probes were generated using DIG RNA Labeling Kit (Roche) and a dRNaseZ full-length cDNA clone, SD27051 (FBcl0284202).

Northern and Western blot analysis

Total RNA from whole animals or dissected tissues was prepared and 5–10 μg of each sample was fractionated either on an agarose (for mRNA analysis) or polyacrylamide gel (for tRNA analysis) for subsequent Northern blotting and hybridization as described earlier (Xie et al., 2011). Sequences of oligonucleotide probes are listed in the supplemental Table S1. Protein extracts were made by homogenizing whole animals directly in Laemmli sample buffer (Bio-Rad). Aliquots were separated on a polyacrylamide gel and subjected to Western blotting and hybridization with the anti-V5 (Invitrogen, 1:10,000) and anti-α-tubulin (Sigma, 1:5,000) antibodies following previously described protocol (Dubrovsky et al., 1996).

Translation assay

The incorporation of ³⁵S-methionine into protein, relative to total protein, was measured in control and knockout animals using a protocol described previously (Hall et al., 2007). Briefly, triplicate sets of 2 larvae were dissected open, inverted halves were incubated with gentle agitation in PBS containing 1 mCi/ml ³⁵S-methionine (MP Biomedicals) for 30 min at room temperature and then washed in cold PBS. Larval tissues were homogenized in lysis buffer (100 mM Tris-Cl, pH 8.0, 100 mM NaCl, 0.5% Triton X-100) and protein extracts were cleared by centrifugation for 5 min at 16,000 g. After extracts were normalized using Bradford assay (Bio-Rad), equal amounts of protein were precipitated with trichloroacetic acid, washed with cold ethanol, air-dried, and dissolved in 100 μL resuspension buffer (62.5 mM Tris-Cl, pH 6.8, 4% SDS, 8M urea). Aliquots of each protein extract were mixed with scintillation cocktail and 1-minute counts were obtained.

Clonal analysis

Mosaic animals were generated with the FRT/FLP system (Golic, 1991). As a marker for homozygous clones, we used ubiquitously expressed nuclear green fluorescent protein GFP^nls (Davis et al., 1995). The RNZ^ED24 mutation was recombined onto FRT-carrying 2^nd chromosome. For somatic clones, we crossed hsFLP¹²;FRT^G13,GFP^nls females to FRT^G13,RNZ^ED24/CyO,GFP males and induced clones in various tissues with a HS treatment at 37 °C as follows: 2 h in embryos (4 ± 1 h AED) for fat body and salivary gland clones, 1 h in first and second instars for imaginal discs, and 1 h in 3–4 day old females for ovaries.

Immunostaining and microscopy

Larvae were dissected at the 3^rd instar wandering stage, and adult females were dissected at 72 h after HS treatment. Tissues were fixed with 4% formaldehyde, washed five times in PBS containing 0.1% Triton X-100 (PTX), blocked for 1 hr in PTX containing 1% BSA (PTXB), and incubated with primary antibodies in PTXB overnight at 4 °C. The following primary antibodies were used, with the dilutions indicated: mouse anti-BrdU (DSHB, 1:100), rabbit anti-PH3 (Upstate, 1:500), mouse anti-V5 (Invitrogen, 1:200). After hybridization, tissues were again washed, blocked, and incubated for 2 hr with secondary antibodies: FITC-goat anti-rabbit (Jackson ImmunoResearch, 1:200), Cy3-goat anti-mouse (Jackson ImmunoResearch, 1:400). Nuclei were counterstained with 1 μg/mL Hoechst 33258 (Sigma) for 10 min. After washes, tissues were mounted in Fluromount-G (SouthernBiotech) and visualized with a Zeiss AxioImager M1 microscope equipped with ApoTome unit. Fluorescent images were captured with AxioCam MRm camera, and color images were captured with AxioCam MRc camera. All images were processed using Adobe Photoshop CS2. Scale bars are shown as indicated.

BrdU incorporation

For BrdU labeling, tissues were dissected and incubated in PBS solution containing 100 μg/mL BrdU for 1 hr (ovaries) or 15 min (imaginal discs). After BrdU incorporation, tissues were fixed as above for immunostaining, incubated in 2N HCl for 30 minutes at room temperature, neutralized with 100 mM Borax and then washed three times in PTX before blocking and hybridizing with anti-BrdU antibody.

TUNEL assay

TUNEL staining was performed with the Fluorescein In Situ Cell Death Detection Kit (Roche) according to the protocol described earlier (McEwen and Peifer, 2005). Imaginal discs dissected from 3^rd instar larvae were fixed and washed in PTX as described for immunostaining. Discs were permeabilized with 0.1 M sodium citrate in PTX for 30 min at 65 °C, rinsed twice in PTX and finally incubated in TUNEL reaction mixture for 2 h at 37 °C. After labeling, discs were washed in PTX and mounted on slides as described above. As a positive control for the procedure, a sample of imaginal discs from wild type animals was treated with DNase I prior to labeling.

Analysis of mutant ovaries and testes

dRNaseZ knockout animals were heat-shocked every 24 h starting from day one AED till adult eclosion. Flies were aged for 7–8 days and then dissected. For phalloidin and Hoechst staining, ovaries were fixed with 4% formaldehyde in PBS for 20 min and stained with 165 nM Texas Red-phalloidin (Sigma) and 1 μg/mL Hoechst 33258 (Sigma) for 20 min. After washing in PBS, samples were mounted in Fluoromount-G and examined under the microscope. For MTRed and Hoechst staining, testes were dissected in PBS, transferred to poly-l-lysine-treated slides, cut open to liberate individual sperm cysts, and gently squashed under a siliconized coverslip. After removing the coverslip in liquid nitrogen, slides were washed with 100% ethanol for 10 min at −20 °C, fixed with 4% formaldehyde for 7 min and stained with 100 nM MTRed (Sigma) and 1 μg/mL Hoechst. After several washes, slides were mounted and examined as described above.

Fluorescence in situ hybridization (FISH)

tRNA in situ hybridization was adapted from previously published protocols (Sarkar and Hopper, 1998; Wilk et al., 2010). Larval tissues were fixed in PBS containing 4% formaldehyde, 0.3% Triton-X 100, and 0.1% picric acid for 20 min at room temperature, and washed three times in PBS containing 0.1% Tween-20 (PBT). Tissues were then permeabilized with 80% acetone at −20 °C for 10 min and post-fixed with 4% formaldehyde in PBT. After washing with PBT, samples were pre-hybridized in 50% Formamide, 4× SSC, 0.1% Tween-20, 50 μg/mL Heparin, 500 μg/mL salmon sperm DNA, and 125 μg/mL E. coli tRNA, for 2 hr at 37 °C. Probes were labeled using DIG Oligonucleotide 3′-end Labeling Kit (Roche) per manufacturer’s instructions. 80 pmol/mL of probe was added to the mix and hybridization was carried out at 37 °C overnight. Tissue samples were washed in 50% Formamide, 4× SSC, 0.1% Tween-20, rinsed in PBT, and then blocked for 1 hr in PBT containing 1% milk powder. Probes were detected with Rhodamine-conjugated mouse anti-DIG antibody (Roche, 1:400). Nuclei were counterstained with 1 μg/mL Hoechst.

Mitotic index

Imaginal wing discs were immunostained with the anti-phosphohistone H3 (PH3) antibody. Images were taken and processed using Zeiss AxioVision Rel. 4.8, Adobe Photoshop CS2, and Image J 1.44o. The mitotic index was calculated as described previously (Martin and Morata, 2006).

Flow cytometry

20–40 3^rd instar wing discs per genotype were dissected in PBS and dissociated by trypsin (Sigma) in EDTA/PBS buffer as described previously (Neufeld et al., 1998). Dissociated cells were washed, and resuspended in 200 μL of cold PBS, fixed by six stepwise additions of 100 μL each of cold 100% ethanol. After overnight 4 °C fixation, cells were resuspended in PBS containing 10 μg/mL Propidium iodide (Invitrogen) and 100 μg/mL RNase A (Roche). DNA amount was analyzed with a BD Accuri C6 flow cytometer. Cell cycle was analyzed with the FlowJo 7.5 software.

RESULTS

Generation of dRNaseZ knockout

To isolate a null allele of the dRNaseZ gene, we employed a technique of transposase-mediated imprecise excision. The P-element was mobilized and potential mutations were recovered by scoring for lethality in combination with Df(2R)12, a large deletion that removes the interval 46F1-47A10 encompassing the dRNaseZ locus. After screening through 100 chromosomes with independently generated excisions, we identified nine lethal mutations. Genomic DNA from homozygote mutants was further characterized by sequencing PCR amplicons spanning deletion endpoints. Three lines – RNZ^ED4, RNZ^ED120, and RNZ^ED24 – had deletions that appear to affect exclusively the dRNaseZ gene (Fig. 1A). The upstream breakpoint of each deletion is at the site of the P-element insertion; the downstream – is in the coding sequence. We selected RNZ^ED24 for further study. It is the longest deletion extending from −11 to +2126-nt position relative to the dRNaseZ transcription start site, thus, removing ~85% of the open reading frame (ORF). Northern blot hybridization did not detect any transcripts indicating that RNZ^ED24 is a null allele (Fig. 1B).

Figure 1. dRNaseZ knockout and rescue.

(A) Genomic map showing two neighboring genes, dRNaseZ and CG12909, and the insertion site of the P-element (open triangle). Arrows mark the transcription start sites. Filled boxes indicate coding regions. A break indicates an intron. Brackets show fragments deleted in RNZ^ED4, RNZ^ED120, and RNZ^ED24. (B) Northern blot analysis of RNA samples prepared from synchronously developing heterozygous RNZ^ED24/+ (WT) and homozygous RNZ^ED24/RNZ^ED24 (KO) siblings 2 days AED. rp49 is a loading control. (C) Viability of WT (white) and KO (grey) animals was calculated as the variance from the initial number of 1^st instar larvae. (D) WT (left) and KO (right) larvae 4 days AED. (E) Morphology of mouth-hooks dissected on day 4 AED indicates that WT (top) siblings are in 3^rd and KO (bottom) are still in the 2^nd instar. (F) Northern blot analysis of RNA samples from WT and KO larvae of day 1 and 2 AED. The internal-probe detects mature tRNA^His; the 3′-probe detects primary transcripts and processing intermediates. (G) Western blot analysis of the V5-tagged dRNaseZ protein, expressed under control of the UAS, native, and HS (0.5, 4, 8, 12, and 16 h after HS treatment) promoter in transgenic animals. α-Tubulin is a loading control.

Homozygous or hemizygous RNZ^ED4 embryos hatch into larvae at the same time and size as heterozygous siblings, suggesting that their embryonic development is normal. However, they do not live long as 80% of them die soon after first larval molt (Fig. 1C). Few animals that survive beyond 3 days AED do not grow and die within the next few days as undersized second instars (Fig. 1D,E).

The KO of dRNaseZ implied that mutant larvae might be defective in tRNA processing. To test this, we performed a Northern blot analysis of five tRNA transcripts (tRNA^His, tRNA^Thr, tRNA_i^Met, tRNA^Trp, and tRNA^Tyr) from control and knockout animals. As shown in Figure 1F, the 3′ probe reveals two bands: a primary transcript of tRNA^His that is present in both mutant and heterozygous siblings, and a processing intermediate with an extension at the 3′ end that is present only in mutant larvae. Probes specific to other tRNAs revealed the same trait – accumulation of processing intermediates in cells of mutant larvae (supplemental Fig. S1). Thus, the loss of dRNaseZ abrogates tRNA 3′ end processing.

If developmental arrest and early larval lethality of RNZ^ED24 mutants are due to the lack of the dRNaseZ protein, the impairment should be rescued by a dRNaseZ-expressing transgene. We prepared three constructs encoding V5-tagged dRNaseZ ORF placed under the control of either UAS, HS, or native promoter. Western blot analysis confirmed that transgenic flies carrying these constructs express ectopic dRNaseZ protein (Fig. 1G). Examination of protein extracts from hs-RNZ transgenics following a single heat pulse revealed, however, a relatively rapid turnover of the ectopic protein (~12 h half life), suggesting that efficient rescue of KO with this construct would require multiple HS pulses delivered every 24 h during development. As shown in Table 1, expression of dRNaseZ protein driven by any of the transgenes can efficiently rescue null mutant flies to adulthood. This test confirms that the only essential gene affected in the RNZ^ED24 line is dRNaseZ.

Table 1.

Rescue of RNZ^ED24 null mutants by the dRNaseZ WT protein expressed via UAS, HS, or genomic transgene.

Genotype a	Viability b	(Cy/Cy+)
RNZED24/RNZED24/CyO;uasRNZ/tubGal4	93%	(339/157)
RNZED24/RNZED24/CyO;uasRNZ/TM3Sb	0%	(267/0)
RNZED24/RNZED24/CyO;genRNZ/+	100%	(272/139)
RNZED24/RNZED24/CyO; +/TM3Sb	0%	(151/0)
RNZED24/RNZED24/CyO;hsRNZ/+	92%	(312/144)
RNZED24/RNZED24/CyO;+/TM3Sb	0%	(194/0)

^aRescue crosses are described in the Material and method section.

^bViability was calculated by comparing the number of the rescued progeny (Cy⁺) to their heterozygous siblings (Cy).

dRNaseZ expression

Previous analysis of dRNaseZ expression by Northern hybridization showed that its transcripts could be detected at all stages, but there was a clear developmental profile: two major peaks of abundance were found in adult females and in embryos (Dubrovsky et al, 2004). These peaks were confirmed by RNA in situ hybridization (supplemental Fig. S2). During oogenesis, high levels of dRNaseZ transcripts accumulate in nurse cells of stage 10 egg chambers and later in mature oocytes and pre-blastoderm embryos. The maternal inheritance of dRNaseZ mRNA explains the survival of RNZ^ED24 null mutants through embryogenesis and first instar.

Using fully functional V5-tagged genomic transgene, we followed a tissue-specific expression of dRNaseZ at the protein level. Consistent with the maternal supply of its transcripts, dRNaseZ was detected at all stages of embryogenesis tracking mitotically dividing cells with a higher level of staining (Fig. 2A–D). During the third larval instar, dRNaseZ is predominantly expressed in still proliferating imaginal discs and CNS (Fig. 2E–H and supplemental Fig. S3), while not so much in endoreplicating tissues such as salivary glands, fat body, and gut. High presence of dRNaseZ protein was detected in the prepupal gonads and in dividing cells of the hindgut imaginal ring and midgut imaginal islands (Fig. 2I,J,K). Abundant levels of dRNaseZ were found in adult fly ovaries (Fig. 2L). It appears first in region 2 of the germarium (Fig. 2L′), but much higher levels are present in the follicular epithelium and nurse cells of egg chambers at stage 2 to 11. Though, no protein is loaded into the oocyte. In testes, dRNaseZ accumulation is high, but restricted to a small area close to the apical tip, where gonialblasts undergo four mitotic divisions yielding 16-cell cysts (Fig. 2M,M′). The level of dRNaseZ rapidly declines through the zone of primary spermatocyte growth and becomes undetectable in the distal part of the testis. Spatial and temporal patterns of dRNaseZ expression suggest that it functions in all cells, but the higher presence in dividing cells implies that dRNaseZ may have an important role in cell proliferation.

Figure 2. Expression pattern of dRNaseZ.

Expression of the dRNaseZ protein was determined by immunostaining with anti-V5 antibody in transgenic animals carrying the genRNZ-V5 construct. Lateral (A–C) and ventral (D) views of whole-mount embryos (anterior to the left). (A) At Stage 11, dRNaseZ is ubiquitously expressed throughout extending germ bands, with higher levels in the gnathal buds (gb) and salivary gland (sg) placodes (bracket). (B) At stage 13, dRNaseZ is expressed in all neuroblasts and myoblasts, with higher levels in the optic lobes (open arrowhead), gb/sg (bracket), hindgut (closed arrow) and Malpighian tubules (open arrow). (C,D) At stage 17, dRNaseZ has higher level of expression in the embryonic longitudinal connectives (closed arrows), ring gland (closed arrowhead), and imaginal primordia (open arrowheads). In late third instar larvae, dRNaseZ is expressed in CNS (E), leg discs (F) wing discs (G), and eye discs (H). In pre-pupae, dRNaseZ is expressed in gonads (I), hindgut imaginal ring (J, closed arrow), and midgut imaginal islands (K). In adults, dRNaseZ is expressed in ovaries (L,L′) and testes (M,M′).

dRNaseZ knockout affects gametogenesis rendering adult flies sterile

Germline expression of dRNaseZ implies a role during gametogenesis. To get a better judgment, we turned to a conditional rescue model, in which a heat shock promoter drives expression of the V5-tagged dRNaseZ transgene. Heat pulses delivered daily allowed a complete rescue of RNZ^ED24 null mutants to adulthood. Adult flies did not require any additional heat pulses for their viability and were kept without further HS. After crossing RNZ^ED24 males and females to the wild type counterparts, we discovered that these flies had low fertility and in few days after eclosion became completely sterile. Returning rescued flies to HS treatment restored their fertility, demonstrating that dRNaseZ depletion is the cause of a malfunction during gametogenesis.

We analyzed defects in gametogenesis of RNZ^ED24 females and males by fluorescence microscopy. After 8 days without dRNaseZ supply, oogenesis in RNZ^ED24 females stalls at stage 8–9 (Fig. 3A). In addition, in mutant egg chambers, we observed extra-numeral germ line cells. A wild type egg chamber has one nascent oocyte connected to 15 nurse cells surrounded by a layer of follicle cells (Fig. 3B-B″,C-C″), while a mutant egg chamber shown in Figure 3(E-E″,F–H) has three oocytes and more than 40 nurse cells of different size. Some mutant egg chambers have the follicle cell layer completely missing or composed of significantly fewer cells (Fig. 3D-D″). Apparently a failing proliferation of follicle cells causes a fusion of several mutant cysts as they bud off from the germarium. Thus, during oogenesis dRNaseZ is required for follicle cell proliferation and nurse cell growth and development.

Figure 3. Defects in dRNaseZ knockout ovaries.

(A) Bright-field images of ovaries from WT (RNZ^ED24/+;hsRNZ) and conditionally rescued KO (RNZ^ED24/RNZ^ED24;hsRNZ) adult females. WT (B-B″,C-C″) and KO (D-D″,E-E″) egg chambers were stained with Texas Red-phalloidin for F-actin (B–E) and Hoechst for DNA (B′–E′). In merge (B″–E″): actin is red, DNA is green. Egg chambers are oriented with anterior to the top. Panels in (B-B″) and (D-D″) are single focal plane images; panels in (C-C″) and (E-E″) are overlay images of several successive focal planes. Panels in (F–H) are three focal planes showing three oocyte nuclei (arrows) in the same Hoechst stained compound egg chamber, as in (E′).

Upon dissection of males, we found that RNZ^ED24 testes look thinner than those of control males (Fig. 4A,D). A quantification assay showed that in the middle section control testes are 93.0±8.5 μm in diameter, and mutant testes are 55.1±6.6 μm (P<0.001). After Hoechst staining, all stages of spermatogenesis were visible in control testes starting from mitotically dividing gonialblasts at the apical tip, cysts with primary spermatocytes and differentiating spermatids in the middle section, and bundles of mature spermatids undergoing individualization in the distal section (Fig. 4B,C,G,H). In contrast, mutant testes displayed several defects. First, the KO of dRNaseZ caused a reduction in the number of mitotically dividing gonialblasts at the apical tip (Fig. 4B,E). Second, all post-meiotic germ cells (e.g. onion stage or elongated spermatids) were absent or underrepresented. After counting cysts of individualizing spermatids that appear as bundles of needle-shaped nuclei (Fig. 4C,F), we found that control testes contain 54±4 and mutant – only 22±3 bundles. Instead, the most abundant cysts of mutant testes were those with 16 fully-grown primary spermatocytes (Fig. 4I-I″). These pre-meiotic cells contain a large nucleus with partially condensed chromosomes (Fig. 4I) and misshapen Nebenkerns (Fig. 4I′) indicating a failure of meiosis. We therefore conclude that during spermatogenesis dRNaseZ is required for mitotic and meiotic germ cell divisions.

Figure 4. Defects in dRNaseZ knockout testes.

(A,D) DIC images of testes from WT (RNZ^ED24/+;hsRNZ) and conditionally rescued KO (RNZ^ED24/RNZ^ED24;hsRNZ) adult males. Panels in (B,C,E,F) are magnified areas indicated as white (B,E) and black (C,F) dashed boxes in (A,D). Brackets in (B,E) indicate the location of mitotically dividing goniablasts. WT (G-G″,H-H″) and KO (I-I″) testis squashes are stained with Hoechst for DNA (G–I) and MTRed for mitochondria (G′–I′). In merge (G″–I″): DNA is green, Mitochondria are red. Dashed boxes indicate magnified areas shown as inserts in the right corner of each panel.

dRNaseZ knockout affects polyploid cell size autonomously

Lack of dRNaseZ affects early larval growth, which in wild-type animals is achieved primarily through the growth of polyploid tissues, such as salivary gland, fat body, muscle, and gut, which comprise the bulk of the larval body. Post-mitotic cells of these tissues grow via endoreplication, a specialized cell cycle when cells increase their ploidy and size but do not divide. To test whether the growth arrest phenotype reflects a cell-autonomous defect in endoreplicative cell growth, we used the FLP/FRT technique to create mosaic larvae with patches of homozygous RNZ^ED24 mutant cells in the fat body (Fig. 5). Somatic clones were induced in early embryogenesis (at 1–4 h AED) and examined in the late larval stages. Mutant cells marked by the absence of GFP were small, with small nuclei (Fig. 5B-B‴), indicating that dRNaseZ is required cell-autonomously to support growth of endoreplicating tissues.

Figure 5. dRNaseZ is essential for endoreplicating cell growth.

Somatic clones in the fat body were generated in wild-type hsFLP;FRT,GFP/FRT,+ (A-A‴) and heterozygous hsFLP;FRT,GFP/FRT,RNZ^ED24 (B-B‴) larvae. Wild type (WT) and RNZ^ED24 (KO) clones are marked by the absence of GFP (outlined with dotted lines). Samples were stained with Hoechst (A′,B′) and Texas Red-phalloidin (A″,B″). In merge (A‴,B‴): GFP is green, DNA is blue, actin is red. Note that KO cells are much smaller than cells in the twin-spot marked with double GFP (B‴), while in WT control corresponding cells have comparable size (A‴).

As growth of larval tissues is mainly driven by endoreplication (Edgar and Orr-Weaver, 2001), we asked whether a small nuclear size in the polyploid cells of somatic clones is indicative of dRNaseZ requirement for DNA replication. To test that we examined if dRNaseZ KO could negate gene amplification in the ovarian follicle cells. Late in oogenesis, several clusters of chorion genes undergo multiple rounds of replication and could be visualized by BrdU incorporation. First, we induced mitotic recombination in heterozygous RNZ^ED24 mutant females, then dissected ovaries were incubated in BrdU, and examined by immunofluorescent microscopy (Fig. 6). Mutant (GFP negative) clones display the same labeling pattern as cells in the twin spot (double GFP) or surrounding heterozygous cells: four BrdU foci in stage 10B (Fig. 6A″) and one strong in stage 11 follicle cells (Fig. 6B″).

Figure 6. dRNaseZ is dispensable for DNA synthesis.

RNZ^ED24 homozygous clones of ovarian follicle cells were generated in heterozygous FRT,GFP/FRT,RNZ^ED24 flies. Mutant cells of the clones are identified by the absence of GFP (marked with yellow dotted lines); wild type cells of the twin-spot clones are labeled with the double dosage of GFP (marked with red dotted lines). Egg chambers of stage 10B (A-A‴) and stage 11 (B-B‴) were stained with Hoechst (A′,B′) and anti-BrdU antibody (A″,B″). In merge (A‴,B‴): DNA is blue, BrdU is red. Gene amplification can be detected by the presence of puncta of BrdU incorporation in both the WT and KO cells.

Thus, dRNaseZ is not required for DNA synthesis per se, rather the small nuclear size observed in somatic clones of endoreplicating tissues is a secondary consequence of growth arrest.

Larval growth arrest and protein synthesis

As tRNA is a vital component of the translation apparatus, we suggested that dRNaseZ is essential for cellular protein synthesis in actively growing and/or proliferating tissues. To test if growth defects of RNZ^ED24 mutants are due to inefficiency of the translational machinery, we examined [³⁵S]-methionine incorporation by larval tissues as a measure of protein synthesis. As homozygous KO mutants are short living and too small for this in vivo analysis, we employed a conditional rescue model. One HS treatment was sufficient to rescue early lethality and developmental arrest. Homozygous animals underwent two larval molts and reached third instar but still failed to attain wild-type size and never pupariate. Instead, they keep crawling in the food for about ten days and then die.

Two groups of conditionally rescued mutants were compared: first received one HS treatment, and second – multiple, every 24 h. As they reached third instar, larvae were dissected, incubated in PBS with [³⁵S]-methionine, and incorporation of radiolabeled amino acid was measured relative to total protein. As an additional control, tissues of wild type larvae were pretreated for 30 min with protein synthesis inhibitor cycloheximide (10 μg/ml) prior to labeling. This produced a reduction in radioactive counts by about 70% relative to control values indicating the efficacy of the assay (Fig. 7). Mutant larvae that received a single HS had a significantly lower level of [³⁵S]-methionine incorporation than control heterozygous animals or homozygous larvae that received multiple HS treatments. These results indicate that dRNaseZ knockout reduces rate of translation and that appears to halt cell growth.

Figure 7. dRNaseZ KO damages protein translation efficiency.

WT (RNZ^ED24/+;hsRNZ) and conditionally rescued KO (RNZ^ED24/RNZ^ED24;hsRNZ) larvae were synchronized at egg deposition, and received HS treatments as indicated. The translation efficiency was assessed as larvae reached 3^rd instar (5 days AED). The values of translation efficiency are normalized to the total protein amount, and shown on the Y-axis as percentage of control (WT larvae treated with 1×HS). CHX: cycloheximide.

dRNaseZ knockout affects the nuclear-cytoplasmic distribution of tRNA

Given that dRNaseZ is a part of the tRNA maturation pathway, it was not surprising to find that the KO of this enzyme lessened the efficiency of cellular protein synthesis. However, a surprising observation was that reduced rate of translation did not reflect the abundance of tRNA molecules. Northern blot analysis revealed similar levels of mature tRNAs in control and knockout animals for isoacceptors from each tRNA family (supplemental Fig. S4). If the pool of mature tRNA molecules is not severely affected, then what is the mechanism of translational deficiency? We suggested that mutant cells respond to the breakdown of tRNA processing by relocating cytoplasmic tRNA to the nucleus, thus, making it unavailable for protein synthesis.

To evaluate nuclear-cytoplasmic distribution of tRNA in wild type and mutant cells, we performed a fluorescence in situ hybridization assay (FISH) on conditionally rescued animals. We chose to study tRNA^Thr, one of most abundant tRNAs, and designed a 35-nucleotide probe to locate mature tRNA^Thr at the subcellular level within big cells of several endoreplicating tissues such as salivary gland, proventriculus (foregut), Malpighian tubules, and body-wall muscle. Heterozygous (control) and homozygous RNZ^ED24 larvae carrying the hs-RNZ transgene received one HS on day 1 AED. Four days later, tissues of control and rescued RNZ^ED24 animals were dissected and processed in parallel. FISH with control tissues generated a signal that was primarily cytoplasmic in salivary gland, proventriculus, and malpighian cells, and more uniform in muscle cells (Fig. 8). Both localization patterns are consistent with tRNA participating in cytoplasmic protein translation and similar to nuclear-cytoplasmic distribution of mature tRNA in cells of budding yeast and rat hepatoma (Shaheen and Hopper, 2005; Shaheen et al., 2007). In contrast, mutant cells displayed a strong nuclear accumulation of tRNA^Thr in all four tissues. Tissues hybridized with a non-specific probe or treated identically without a probe did not generate any fluorescence (data not shown), indicating that hybridization signals were specific. The nuclear-cytoplasmic re-distribution of tRNA^Thr molecules in dRNaseZ KO cells was confirmed with a quantification assay, when we determined the ratio between nuclear and cytoplasmic signals on a population of cells derived from tissues hybridized in several independent experiments (Fig. 8). We also extended our FISH assay by designing and hybridizing probes for tRNA_i^Met and tRNA^His, both of which confirmed our initial observation – homozygous RNZ^ED24 cells displayed a strong nuclear accumulation of tRNA molecules (supplemental Fig. S5).

Figure 8. dRNaseZ KO results in tRNA^Thr nuclear accumulation in different tissue cell type.

Left panels: salivary gland, proventriculus, malpighian tubule and body wall muscle dissected from 3^rd instar WT (RNZ^ED24/+;hsRNZ) and conditionally rescued KO (RNZ^ED24/RNZ^ED24; hsRNZ) larvae. The subcellular distribution of tRNA^Thr was monitored by FISH (Cy3) using DIG-labeled oligonucleotide probe. The nuclei were counterstained with Hoechst. Right panels: results of a quantification assay of the nuclear-cytoplasmic tRNA distribution obtained on a population of cells from three independent FISH experiments. At least 60 cells were tested for each tissue. The X-axis shows groups of cells with a specific nucleus-to-cytoplasm signal ratio. The Y-axis shows the number of cells in each group, as a percentage of cells being tested.

dRNaseZ knockout affects imaginal disc growth autonomously

Growth arrest of dRNaseZ KO could be partially rescued with one or two heat pulses delivered to larvae carrying the hs-RNZ construct. By day 5 AED, mutant larvae grew to a size close to third instar (Fig. 9A–D). However, upon dissection we found that only endoreplicating tissues increased in size during this rescue period, mitotic tissues failed to reach wild type size (compare panels B′–D′ and B″–D″ in Fig. 9). To test whether dRNaseZ requirement in mitotic tissues is cell autonomous, we utilized FLP/FRT-mediated recombination to generate RNZ^ED24 mutant cells in imaginal discs. When recombination was induced early at the beginning of the 1^st instar (96 h before dissection), no RNZ^ED24 mutant clones (GFP negative) were found, while large wild type twin spots were easily identifiable by double dosage of GFP (supplemental Fig. S6). Apparently, cells lacking dRNaseZ are outcompeted by faster growing neighbors and eliminated from the imaginal epithelium. When recombination was induced later in development (60–70 h before dissection), each wild type twin spot was accompanied by an RNZ^ED24 mutant clone (Fig. 10). Importantly, mutant clones were substantially smaller than their wild-type twins. These observations indicate that the absence of dRNaseZ protein causes an autonomous growth disadvantage in the developing imaginal discs either because of reduced cell proliferation or viability.

Figure 9. Growth of endoreplicating and mitotic tissue in dRNaseZ KO larvae.

(A–D) Shown are WT (RNZ^ED24/+;hsRNZ) and KO (RNZ^ED24/RNZ^ED24;hsRNZ) larvae 5 days AED. The KO larvae were raised as three separate groups: (A) one group was continuously kept at 25 °C, (B) the other group received one HS treatment (1×), and (C) the third one – two HS treatments (2×). Fat bodies (A′–D′) and imaginal discs (B″–D″) were dissected 5 days AED and stained with Hoechst and Phalloidin.

Figure 10. dRNaseZ is essential for imaginal disc growth.

Somatic clones in the eye imaginal discs were generated in wild-type hsFLP;FRT,GFP/FRT,+ (A-A″) and heterozygous hsFLP;FRT,GFP/FRT,RNZ^ED24 (B-B″) larvae. Wild type (WT) and RNZ^ED24 (KO) clones are identified by the absence of GFP (marked with yellow lines). Discs were stained with Hoechst (A′,B′). In merge (A″,B″): GFP is green, DNA is red. Note that KO clones are much smaller than their respective twin-spot clones (B″), while in WT control corresponding clones have comparable sizes (A″).

dRNaseZ knockout decreases cell proliferation but not cell survival

Analysis of tissues from RNZ^ED24 mutants, rescued with two pulses of hs-RNZ ectopic expression, showed that despite reaching wandering stage mutant larvae had imaginal discs that failed to grow to full 3^rd instar size (Fig. 11B-B″). Reduced growth of imaginal tissues could be due to defects in cell proliferation or survival. To exclude low survival alternative, we used the TUNEL Kit to detect DNA fragmentation, which is indicative of cell death. Wing discs dissected from rescued RNZ^ED24 larvae were smaller (Fig. 11D), but did not show elevated levels of cell death (Fig. 11D′) when compared to discs dissected from heterozygous siblings (Fig. 11C,C′). Thus, loss of dRNaseZ does not affect cell survival suggesting that growth deficit of mutant discs may result from inefficient cell proliferation.

Figure 11. dRNaseZ KO affects imaginal disc growth but does not cause cell death.

Shown are wing imaginal discs of WT (RNZ^ED24/+;hsRNZ) and KO (RNZ^ED24/RNZ^ED24;hsRNZ) larvae rescued with two (2×) HS treatments. Upper panel: discs were dissected on 4^th, 5^th, and 6^th day AED, and stained with Hoechst. Note that even after two HS pulses mutant discs are still smaller than WT (compare A and B-B″). Lower panel: wing discs of WT (C,C′) and KO (D,D′) larvae rescued with 2×HS were stained with Hoechst (C,D) and TUNEL (C′,D′).

To assess cell division in RNZ^ED24 mutant wing discs, we carried out BrdU pulse labeling to detect DNA replication and anti-PH3 staining to identify cells in mitosis. There was significantly less BrdU incorporation in mutant discs (Fig. 12A,B), indicating that knockout of dRNaseZ delays cell cycle progression with less cells in the S phase. To our surprise, we found numerous cells in mitosis revealed in mutant discs with anti-PH3 (Fig. 12A′,B′). The mitotic index calculated in a control disc was 1.6±0.2 (n=6), whereas in a mutant disc it was 2.8±0.2 (n=5). These data suggest that dRNaseZ KO delays the cell cycle progression at the G2/M transition.

Figure 12. dRNaseZ KO affects cell cycle progression.

(A,A′) and (B,B′) panels show 3^rd instar wing imaginal discs dissected from WT (RNZ^ED24/+;hsRNZ) and 2×HS rescued KO (RNZ^ED24/RNZ^ED24;hsRNZ) larvae, respectively. Discs were stained for BrdU (A,B) and PH3 (A′,B′). (C) FACS analysis of cells from WT (black) and KO wing discs dissected from 2×HS larvae on 5^th (dark gray) and 8^th (light gray) day AED. Histograms display DNA content (X-axis) and cell number (Y-axis). Inset shows percentage distribution of cells between G1, S, and G2-M phases in each group tested. (D) Forward scatter (FSC) analysis of the same cell samples as in (C). WT cells are in black, KO cells are in dark gray (5^th day AED) and light gray (8^th day AED).

To directly test if dRNaseZ function is required for progression from the G2 into M phase, we used flow cytometry to examine the cell cycle profiles of wild type and RNZ^ED24 mutant cells in precisely staged 3rd instar wing discs. In Figure 12C, the control profile taken on day 5 AED shows a typical distribution of cells between three fractions representing the G1, S, and G2/M phases (Neufeld et al., 1998). However, the KO profiles taken on day 5 and 8 AED show a clear shift with a higher mutant cell count in the G2/M fraction at the expense of cells in G1 and S, once again indicating a G2/M cell cycle delay. Furthermore, analysis of the forward scatter profile (Fig. 12D), as a relative measure of cells size, shows that, on average, cells lacking dRNaseZ activity become progressively larger than wild type cells by about 7% and 19% on day 5 and 8 AED, respectively. These results indicate that the primary consequence of dRNaseZ KO in mitotic cells is a defect in cell cycle progression rather than cell growth.

DISCUSSION

dRNaseZ and fly development

Our goal was to manipulate the zygotic supply of dRNaseZ to understand its critical role during Drosophila development. To initiate this analysis we generated a knockout allele and studied mutant phenotypes with the conditional rescue system at the organismal and cellular levels. By analyzing animal viability we found that dRNaseZ activity is required from the very beginning of larval development. Maternal supply of dRNaseZ appears to be sufficient to support development of the embryo, however most of RNZ^ED24 homozygous siblings die 48 h after hatching without much growth. Lethality and growth arrest originate exclusively from the loss of dRNaseZ function, as both phenotypes can be fully rescued with a continuous supply of ectopic dRNaseZ from inducible transgenes or a genomic fragment containing a wild-type copy of the gene. Thus, dRNaseZ is required for larval growth, which is consistent with observations on other genes related to protein synthesis and whose knockout caused growth arrest and larval lethality (Galloni and Edgar, 1999; Lachance et al, 2002; Grewal et al., 2007). Drosophila larva is composed of mitotic and endoreplicating tissues, the latter comprising the bulk of larval body. We found that two types of tissues have different requirements for dRNaseZ. One pulse of hs-RNZ expression was enough to rescue partial growth of salivary glands, fat body and gut. However, imaginal discs remained rudimentary or were completely missing, as they were hard to identify in homozygous mutant larvae conditionally rescued with one HS treatment. Apparently a continuous supply of dRNaseZ is needed in mitotic tissues, whereas endoreplicating tissues once received a pulse of ectopic dRNaseZ could grow in a more independent mode. This conclusion is consistent with the expression pattern of dRNaseZ, which is abundant in all mitotically active tissues and present at a much lower level in polyploid larval tissues. The requirement for dRNaseZ in pupae seems to follow a similar pattern, as successful rescue of homozygous RNZ^ED24 adult flies required HS treatment through early pupal development, a period of growth and proliferation of histoblasts, imaginal precursors of the abdominal epidermis and midgut. However, rescue is not affected by the cessation of heat treatment during late pupal and adult stages. This is consistent with our previous observation that dRNaseZ is not necessary for the differentiation and/or maintenance of postmitotic cells (Xie et al., 2011).

While viability of adult flies does not depend on dRNaseZ, its continuous supply is essential for gametogenesis in both males and females. Ovaries of RNZ^ED24 females were composed of defective egg chambers containing multiple germ line cysts that apparently fused together because of failed proliferation of follicle cells. In males the damage was two-fold, as dRNaseZ KO prevented mitosis of gonialblasts and meiosis of primary spermatocytes. Thus, defects in gametogenesis again suggest that dRNaseZ is important for cell division.

dRNaseZ and cell growth

Our study indicates that in endoreplicating tissues dRNaseZ function is important for the increase of cell size. Somatic clones generated in fat body showed that RNZ^ED24/RNZ^ED24 cells were significantly smaller and contained less DNA compared to neighboring heterozygous and homozygous wild-type cells. This result corroborates with our previous finding that tissue-specific dRNaseZ-RNAi caused a dramatic reduction in cell and nuclear size in salivary glands (Xie et al., 2011). As dRNaseZ knockdown impaired tRNA maturation and possibly lowered protein production, we thought that its activity is likely to be directly linked to growth, while small nuclear size might be secondary to the defect in cellular growth. Here we show that homozygous RNZ^ED24 follicle cells display normal DNA amplification, thus indicating that indeed dRNaseZ is dispensable for DNA synthesis. We also show that dRNaseZ KO decreases the rate of bulk protein synthesis by at least 50%. Given that tRNA is an essential component of the translation machinery and the knockout of dRNaseZ completely blocks maturation pathway of tRNA, the latter observation was expected. However, it was a surprise to find out that the loss of dRNaseZ diminished protein synthesis without diminishing the pool of tRNA molecules. When we tested dRNaseZ KO larvae for the presence of mature tRNA from each isoacceptor family, we did not see any significant damage – the abundance of all tRNAs was close to or at the same level as in control animals. Trying to resolve an apparent discrepancy, we proposed that mutant animals experience a deficiency in protein synthesis because tRNAs, depicted in Northern blots as being abundant, might not be fully available for the cellular translation machinery.

To serve its major purpose of delivering amino acids to ribosomes, tRNA must be located in the cytoplasm. At the same time, a number of studies showed that during its lifetime mature tRNA may go through multiple rounds of nuclear export, import, and re-export (Rubio and Hopper, 2011). The shuttling of tRNA between the nucleus and the cytoplasm appears to be conserved as it was found in both yeast and mammalian cell cultures (Shaheen et al, 2007; Takano et al, 2005; Huynh et al, 2010). In cells growing under normal conditions, flows of tRNA in and out of the nucleus appear to be balanced producing a certain nuclear-cytoplasmic tRNA ratio. However, recent studies found that particular external or internal conditions could tilt the balance. For example, genomic DNA damage or nutrient deprivation both affect nuclear-cytoplasmic tRNA dynamics (Shaheen and Hopper, 2005; Ghavidel et al, 2007). In yeast, some mutations damaging tRNA processing or transport lead to the nuclear retention of mature tRNAs (Grosshans et al, 2000; Murthi et al, 2010). We suggested that Drosophila cells could respond to dRNaseZ knockout by relocating the pool of mature tRNA to the nucleus. In FISH experiments, we indeed found that cells without dRNaseZ activity display tRNA nuclear accumulation. All other studies showing nuclear tRNA import were conducted in yeast and cell culture models. This is the first report demonstrating in vivo nuclear accumulation of previously cytoplasmic tRNA in metazoan cells of multiple tissue types. This observation is consistent with reduced translational capacity of dRNaseZ KO tissues and inability of mutant larvae to gain full size of the third instar after a single pulse of HS-rescue.

While the process of tRNA nuclear relocation appears to be conserved in yeast, mammalian, and now insect cells, the biological function of the phenomenon is unknown. Two nonexclusive hypotheses are under consideration – mature tRNA is brought back to the nucleus to go through quality control, and/or to “hide” from cytoplasmic translation machinery under adverse conditions (Rubio and Hopper, 2011). The latter idea is consistent with our observations that cells deficient in dRNaseZ activity relocate tRNA to the nucleus, reduce protein synthesis, stop growing, stop replicating DNA. This response apparently allows mutant cells to avoid apoptosis and survive effectively on a minimal supply of dRNaseZ. Homozygous RNZ^ED24 larvae could live for more than two weeks after just one pulse of HS-rescue, though they never grow to full wild-type size. Thus, tRNA nuclear import could serve as a mechanism to reduce protein translation and improve survival under certain external or internal conditions. How dRNaseZ KO leads to nuclear accumulation of mature tRNA is not clear, as it could be a result of decreased re-export, increased import, or a combination of both.

dRNaseZ and cell proliferation

In imaginal discs that get bigger during larval development via mitotic cell proliferation, dRNaseZ KO autonomously affects tissue growth by slowing down cell cycle progression. It would seem that dRNaseZ KO might affect cell proliferation indirectly by reducing the pool of cytoplasmic tRNAs and thereby diminishing the capacity of the cellular translation machinery. As cell growth and proliferation are tightly coupled to each other and to protein synthesis, this would explain the dependence of imaginal disc growth on dRNaseZ activity. This possibility is consistent with previous observations that inadequate supply or improper maintenance of components of translation machinery such as ribosomes stunts cell growth and proliferation (Thomas, 2000). In Drosophila, elevated tRNA synthesis promotes organismal growth (Rideout et al., 2012; Marshall et al., 2012). The long known Minute mutations affecting ribosomal protein genes manifest in a developmental delay, short thin bristles, poor fertility and viability, which appear to result from reduced cell growth and division (Marygold et al, 2007). Genetic damages to rRNA or translation initiation factors retard cell cycle progression and produce a larval growth arrest phenotype (Galloni and Edgar, 1999; Lachance et al, 2002; Grewal et al., 2007). Similarly, in yeast and mammalian cells, inhibition of rRNA synthesis derails cell growth and cell cycle progression (Bernstein et al, 2007; Donati et al, 2011). It appears that defective components of the translation apparatus affect cell cycle kinetics by preventing the G₁-S phase transition. It is well established that a checkpoint at the end of G₁ commits cells to divide and as such it is particularly sensitive to protein synthesis. Even partial inhibition of protein synthesis with a non-lethal dose of cycloheximide can cause cell cycle blockade at the G₁-S boundary (Liu et al., 2010). Surprisingly, in our study we found that although dRNaseZ knockout reduced cellular protein translation, it inhibited cell proliferation by blocking the G₂-M phase transition. While mutant imaginal discs did not grow or incorporate significant amounts of BrdU, immunostaining with anti-PH3 revealed numerous cells in mitosis. FACS analysis further confirmed that mutant discs were composed of larger cells with G₂ DNA content, suggesting that loss of RNase Z affects cell division more than cell growth. A similar phenotype was observed with some Minute alleles (Marygold et al., 2005, Martin-Castellanos and Edgar, 2002). All together our results indicate that depletion of dRNaseZ lead to cell cycle arrest at the G₂-M transition. Thus, it appears that a mechanism other than reduced protein synthesis makes cell cycle reliant on dRNaseZ activity.

Studies in mammalian cells suggest that RNase Z^L could have functions independent of its enzymatic activity. Human RNase Z^L was shown to interact with cytoplasmic γ-tubulin, which is required for the mitotic spindle formation. Interestingly, ectopic RNase Z^L could obstruct the function of mitotic apparatus and delay cell cycle at G₂-M (Korver et al, 2003).

Another study identified RNase Z^L as a transcriptional scaffold protein in the TGF-β signaling pathway that regulates growth and proliferation of prostate cells. For quite some time, human ELAC2/RNase Z^L has been connected with the occurrence of Prostate Cancer (PCA), though the mechanism is still unclear (Tavtigian et al, 2001). It turned out that in prostatic epithelial cells RNase Z^L can interact with activated Smad2 and other transcription factors thus modulating expression of the TGF-β target genes (Noda et al., 2006). In Drosophila, Decapentaplegic (Dpp)/TGF-β signaling is one of the major pathways that regulates growth of the imaginal disc epithelia (Martin-Castellanos and Edgar, 2002; Affolter and Basler, 2007). Dpp represses the transcription factor Brinker (Brk) and thereby induces expression of Myc that promotes tissue growth (Doumpas et al, 2013). Like its human homolog, dRNaseZ could be a part of the Dpp signaling pathway, then its KO would inhibit signal transduction and arrest growth. Interestingly, dRNaseZ and Myc KO both impair cellular growth and proliferation producing the same phenotype in imaginal discs: cells without dRNaseZ (Fig. 9) or Myc activity form tiny clones that are hard to recover as they get eliminated by cell competition (Pierce et al, 2008; Neto-Silva et al, 2010).

There is a possibility that growth regulation function of dRNaseZ originates from its endonucleolytic activity. Little is known about mechanisms and/or factors that coordinate synthesis and processing of tRNA and growth control, which makes the discovery of a novel class of small regulatory RNAs derived from transfer RNA even more exciting (Pederson, 2010). For our study of particular interest is a series of tRNA-derived RNA fragments (tRF) produced by RNase Z^L endonucleolytic cut off of the 3′ trailer during pre-tRNA processing. Once considered as by-products of tRNA maturation destined for degradation, these short RNAs are viewed now as molecules with specific expression and function. One of these tRFs, tRF-1001, is highly abundant in many cancer cell lines especially in those with high proliferation rate. The knockdown of tRF-1001 caused cell cycle arrest with the specific accumulation of cells in G₂ (Lee et al, 2009). Even though tRFs have not been found yet in Drosophila, a preliminary computational analysis based on pre-tRNA length distribution shows that 24% of tRNA 3′-trailers represent potential tRFs (data not shown). dRNaseZ could modulate cell proliferation rate by regulating levels of these tRFs, although further work is obviously needed to test this idea.

• Drosophila RNase Z^L (dRNaseZ) provides a non-redundant essential function

• dRNaseZ KO affects processing but not the abundance of mature tRNA molecules

• Without dRNaseZ cells relocate tRNA into the nucleus and cut protein synthesis

• dRNaseZ is required for cell growth in endoreplicating tissues

• dRNaseZ is required for cell cycle progression in mitotic tissues