Evidence for a novel function of Awd in maintenance of genomic stability

The abnormal wing discs (awd) gene encodes the Drosophila homolog of NME1/NME2 metastasis suppressor genes. Awd acts in multiple tissues where its function is critical in establishing and maintaining epithelial integrity. Here, we analysed awd gene function in Drosophila epithelial cells using transgene-mediated RNA interference and genetic mosaic analysis. We show that awd knockdown in larval wing disc epithelium leads to chromosomal instability (CIN) and induces apoptosis mediated by activation of c-Jun N-terminal kinase. Forced maintenance of Awd depleted cells, by expressing the cell death inhibitor p35, downregulates atypical protein kinase C and DE-Cadherin. Consistent with their loss of cell polarity and enhanced level of matrix metalloproteinase 1, cells delaminate from wing disc epithelium. Furthermore, the DNA content profile of these cells indicates that they are aneuploid. Overall, our data demonstrate a novel function for awd in maintenance of genomic stability. Our results are consistent with other studies reporting that NME1 down-regulation induces CIN in human cell lines and suggest that Drosophila model could be successfully used to study in vivo the impact of NME/Awd - induced genomic instability on tumour development and metastasis formation.


Results
Down-regulation of awd leads to genomic instability and cell death. We have analysed the effects of Awd depletion in larval wing discs since these primordia are an excellent model system to study CIN and tumourigenesis 17 . As shown in Fig. 1A, the Awd protein is expressed throughout the wing disc 13 (n = 30). We have down-regulated awd expression through UAS/Gal4-driven RNA interference. The engrailed-Gal4 (en) driver has been used to induce expression of the UAS-awd-RNAi (awdi) construct in the posterior compartment cells of larval wing disc (Fig. 1A). The use of the compartment specific en-Gal4 driver allows direct comparison of wild type cells in the anterior compartment versus awd mutant cells in the posterior compartment within a single imaginal disc. Co-expression of the GFP marker allows easy recognition of the domain targeted for awd silencing in en > GFP, awdi larvae (hereafter referred as en > awdi). To mark the anterior domain, we have stained wing discs for cubitus interruptus (ci), whose expression is repressed in the posterior domain by the en gene product 18 . The analysis of en > awdi wing discs has shown patches of cells lacking GFP expression and expressing Awd and Ci protein despite their localisation in the posterior region of the disc where Awd is downregulated (73,3% wing discs; n = 30) (Fig. 1B). Thus, a genomic instability event has involved the second chromosome hosting the en driver leading to impaired en driver activity and loss of heterozygosity of the UAS-GFP and UAS-awdi transgenes. Furthermore, expression of the Ci anterior marker in these GFP − , Awd + cells located in the posterior wing disc domain, demonstrates impaired en gene function. Thus, down-regulation of awd expression leads to genomic instability events that involve both copies of the second chromosome and cause loss of en gene function.
Further analyses of en > awdi wing discs have shown that down-regulation of awd in posterior domain activates JNK as indicated by the presence of the phosphorylated form of JNK (pJNK) (Fig. 1C, n = 35) and by the elevated level of matrix metalloproteinase 1 (MMP1) (Fig. 1D, n = 39), a downstream effector of JNK pathway 19 . Furthermore, these posterior cells exhibit an increased level of apoptotic death as shown by cleaved-Caspase 3 (Casp3) staining (Fig. 1E, n = 30). Genomic instability often leads to CIN and aneuploidy 4,5 . Interestingly, CIN induced by silencing of SAC genes, or downregulation of genes involved in spindle assembly and cytokinesis triggers JNK pathway activation, upregulation of MMP1 and apoptotic cell death 8,9 . Thus, our data suggest that Awd depletion leads to CIN.
Our previous studies on awd function already showed that this gene is required during development for normal wing development since adult flies, mosaic for awd J2A4 loss of function allele, show altered wing morphology 13 . The analysis of en > awdi adults has shown that, in comparison with wild type (Fig. 1F), en > awdi wing discs develop into adult wings with a reduced posterior domain and bent toward the posterior (Fig. 1G, n = 20). This phenotype is consistent with the occurrence of apoptotic cell death induced by Awd depletion in larval wing disc.
awd knockdown leads to abnormal eye differentiation. We extended our analysis of awd gene function to the development of adult eye. The detailed analysis of the eye of adult flies expressing awdi in the larval eye disc under the control of the eyeless-Gal4 driver 20 (ey > awdi) has showed aberrant morphological phenotype. In wild type adult eye about 700 ommatidia, each with an inter-ommatidial bristle, are organised in a highly regular array ( Fig. 2A). On the contrary, ey > awdi adult eyes show clear defects in alignment of ommatidial facets and bristles and with some ommatidia missing (Fig. 2B). Furthermore, we have analysed the effects of awd J2A4 null mutation on eye development. Since this allele is lethal in homozygous condition, we have applied the directed mosaic technique to induce awd J2A4 mosaic clones in the eye disc 21 . We have used the ey-flp line 22 to target FRT/ Flp site specific recombination to the eye disc. Morphological analysis of ey-flp-induced awd J2A4 mosaic adult eyes also shows that a rough eye phenotype results from loss of awd function (Fig. 2C). Overall our results show that knockdown as well as complete loss of awd function in larval wing and eye discs impairs normal development of the corresponding adult structures. Altered differentiation of adult structures arises from CIN induced aneuploidy in wing and eye discs 8 , so our results on the role of awd in maintenance of genomic stability suggest that Awd depletion leads to CIN. awd silencing leads to aneuploidy. To avoid clearing of awd depleted cells, and better characterise the CIN events occurring in those cells, we have co-expressed awdi and p35 under control of en driver. Block of apoptotic machinery through expression of the effector caspase inhibitor p35 23 leads in fact to persistence of CIN cells in tissues. These cells show altered plasma membrane polarity, defects in adherens junctions that connect cells and separate apical and basolateral cellular domains, and delaminate from the epithelium 8 . Then, we have analysed the effects of awdi and p35 co-expression on polarity of wing disc cells (hereafter referred as en > awdi, p35). We have focused on two proteins, the DE-cadherin (DE-Cad) component of adherens junctions and the atypical protein kinase C (aPKC) component of a sub-apical protein complex 24 . Wing discs were dissected from en > awdi, p35 larvae and in 10 out of 16 discs analysed we detected a visible reduction of DE-Cad (Fig. 3A) and aPKC levels (Fig. 3B). The differences in mean fluorescence intensity between GFP negative and GFP positive cells in each of the 16 wing discs have been analysed using a paired t-test. Applying this test the reduction of De-Cad and aPKC stainings in GFP positive cells is statically significant (**p = 0.0081, for DE-Cad and *p = 0.0158, for aPKC) (Supplementary Data S1). Furthermore, en > awdi, p35 wing discs show strong up-regulation of MMP1 expression (Fig. 3C, n = 20). High levels of MMP1 lead to basement membrane degradation and cellular invasiveness in normal and tumoural tissues 19,25 . Interestingly, scanning across the vertical axis (x-z section in Fig. 3C) the posterior region of en > awdi, p35 wing discs shows that cells with high MMP1 level localise on the basal side of the epithelium. Thus, persistence of awd depleted cells in wing disc leads to CIN biological responses including delamination from the epithelium. CIN consists of gain or loss of whole chromosomes or chromosomal regions leading to cellular aneuploidy. Thus, we have analysed the DNA content profile of cells from the posterior compartment of en > awdi, p35 wing discs using as control cells from the anterior compartment (Fig. 3D). Up to 38.5% of awdi cells has DNA content higher than 4n while only 11.3% control cells has a similar DNA content. Since it has been shown that wing disc cells expressing p35 have a percentage of cells with DNA content similar to wild type cells 8 , these results clearly show that awdi cells are aneuploid.

Discussion
We report here, for the first time, a novel role for the Awd endocytic mediator in maintenance of genomic stability in larval disc epithelium. Our results show that depletion of Awd triggers JNK-mediated cell death of wing disc cells and that blocking the cell death machinery results in aneuploidy and cell delamination without overt hyperproliferative effect (Fig. 4). Overgrowth of wing disc hosting aneuploid cells is due to activation of the JNK pathway that promotes expression of Wingless (Wg) upon block of apoptotic cell death. Wg is a mitogenic molecule required in the imaginal discs for growth and patterning 26 and its expression in the aneuploid, delaminating CIN cells triggers growth of neighbouring non-delaminating cells 8 . However, awd J2A4 mutant wing disc cells do not express Wg as consequence of faulty Notch signalling 13 therefore, they cannot promote hyperplasia of the surrounding tissue. Furthermore, lack of hyperproliferation is also observed when aneuploid condition arises from impaired activity of genes controlling karyokinesis. The diaphanous gene (dia) codes for an actin-regulatory molecule which is required during acto-myosin driven contraction of metaphase furrows 27 . Simultaneous depletion of dia gene expression and block of apoptosis do not lead to hyperplastic growth probably due to defective karyokinesis 8 . Intriguingly, Awd is a microtubule-associated nucleoside diphosphate kinase that converts GDP to GTP and the analysis of awd mutant larval brain showed mitotic defects correlated with defective microtubule polymerisation 28 . This raises the possibility that the Awd kinase function plays a role in GTP supply to protein such as Orbit which are required for stabilisation of spindle microtubules 29 .
Two lines of evidence further support the hypothesis that Awd could be involved in karyokinesis. The first comes from studies showing that endosome trafficking and transport to the intercellular bridges of dividing cells plays a critical role during abscission, the last step of karyokinesis 30 . In addition, remodelling of plasma membrane that underlies nuclear divisions in syncytial embryo and cellularisation also requires endocytosis 31,32 . Embryo cellularisation requires the dynamin encoded by the shibire (shi) locus and Rab5 GTPase function since loss of function of either genes arrests ingression of metaphase furrows 33 . Awd functionally interacts with shi locus 12 and it is also required for Rab5 function in early endosomes 13,14 . Thus, a possible role for Awd in cytokinesis should be considered. The second line of evidence comes from studies on NME1, the human homolog of awd gene. This metastasis suppressor gene 34 shares about 78% of aminoacid identity with the awd gene 28 . Down-regulation of NME1 gene expression in diploid cells results in cytokinesis failure and leads to tetraploidy 16 . Our in vivo results show that Awd plays a role in maintenance of genomic stability confirming the high degree of conservation between NME1 and Awd proteins. Drosophila studies have already been crucial in identification of NME1 function in epithelial morphogenesis and our present work shows that it can be a useful model to investigate also this function and its impact on tumour development and progression.

Statistical analyses.
Fluorescence stains have been captured on TCS SL Leica confocal microscope. Five stacks per wing disc have been acquired with xzy scaling of 630 μm. Fluorescence intensity quantification of projected z-stack images has been restricted to neighbour areas of same size in anterior and posterior compartments. Image J software has been used for measurement of fluorescence intensity. Differences between mean fluorescence intensity have been estimated for statistical significance using a two-tailed distribution paired Student's t-test with Prism software.
Scanning Electron Microscopy. ey-Gal4/UAS-awdRNAi and ey-FLP; Act-Gal4, UAS-GFP; awd J2A4 , FRT82B/Tub-Gal80, FRT82B and control (ey-Gal4/UAS-GFP and ey-FLP; Act-Gal4, UAS-GFP; Tub-Gal80, FRT82B) flies were collected and washed several times in water and then dehydrated in 100% ethanol. Flies were then incubated in a solution of ethanol:tetramethylorthosilicate (TMOS) 1:1 for 2 hours and then let dry in TMOS 100% overnight at room temperature in a fume hood. The day after the heads were dissected and carefully mounted on an aluminium stub previously prepared with a double stick carbon tape. Samples were then coated with a gold film before SEM examination. Samples were analysed with SEM JEOL JSM-5400 microscope and images were recorded at accelerating voltage of 15 kV. Mounting of adult wings. The left wings from female flies were washed in 1xPBS, dehydrated in ethanol 100% and then dissected and mounted on glasses in lactic acid/ethanol (6:5). Wing images were captured using a Nikon Eclipse 90i microscope.

Flow cytometry analysis.
Approximately 60 L3 wing discs for genotype were dissected in cold 1xPBS, centrifuged (5 minutes 3000rcf at RT) and dissociated into single cells in 0,05% Trypsin-EDTA (1X) (Gibco) for 4 h in RT. After dissociation, samples were incubated overnight at −20 °C in a solution of ethanol:PBS 2.5:1. After several washes in 1xPBS+EDTA, samples were incubated with propidium iodide, PI (15 minutes) and analysed by FACS.
PI fluorescence was determined by flow cytometry using a Fluorescent activated cell sorter BD FACSaria Cell Sorter (BD Bioscences). Excitation of the sample was carried out using a Coherent Sapphire Solid State laser. Excitation with 488 nm allowed the acquisition of forward-scatter (FS), side-scatter (SS), fluorescence from GFP. Doublets were discriminated using an integral/peak dot plot of PI fluorescence. Optical alignment was based on optimized signal from specialised fluorescent 6 μm particle (BD AccuDrop beads). DNA analysis on single fluorescence histograms was done using BD FACSDiVa software (BD Biosciences).