|Home | About | Journals | Submit | Contact Us | Français|
Myc oncoproteins are essential regulators of the growth and proliferation of mammalian cells. In Drosophila the single ortholog of Myc (dMyc), encoded by the dm gene, influences organismal size and the growth of both mitotic and endoreplicating cells. A null mutation in dm results in attenuated endoreplication and growth arrest early in larval development. Drosophila also contains a single ortholog of the mammalian Mad/Mnt transcriptional repressor proteins (dMnt), which is thought to antagonize dMyc function. Here we show that animals lacking both dMyc and dMnt display increased viability and grow significantly larger and develop further than dMyc single mutants. We observe increased endoreplication and growth of larval tissues in these double mutants and disproportionate growth of the imaginal discs. Gene expression analysis indicates that loss of dMyc leads to decreased expression of genes required for ribosome biogenesis and protein synthesis. The additional loss of dMnt partially rescues expression of a small number of dMyc and dMnt genes that are primarily involved in rRNA synthesis and processing. Our results indicate that dMnt repression is normally overridden by dMyc activation during larval development. Therefore the severity of the dm null phenotype is likely due to unopposed repression by dMnt on a subset of genes critical for cell and organismal growth. Surprisingly, considerable growth and development can occur in the absence of both dMyc and dMnt.
Throughout evolution, biological systems have employed molecular antagonism as a means of maintaining highly regulated and robust control over biochemical reactions, signal transduction pathways, and transcriptional networks (Gerhart and Kirschner, 1997). At the level of transcriptional control there are a number of well documented examples of transcriptional activators and repressors whose mutually antagonistic behavior at specific promoters serve to determine the rate of transcription and the temporal response to signaling (for review see (Barolo and Posakony, 2002)). An interesting case of transcriptional antagonism is provided by the Max transcription factor network, a molecular module comprised of a group of basic-helix-loop-helix-leucine zipper (bHLHZ) transcription factors, all of which form individual heterodimers with the small bHLHZ protein Max. The Max network encompasses the functions of the Myc oncoprotein family and its antagonists, the Mxd family of proteins (for reviews see (Eisenman, 2006; Grandori et al., 2000; Luscher, 2001; Oster et al., 2002).
In vertebrates the expression of Myc family proteins (c-, N-, L-Myc) is induced and maintained in response to a wide range of growth and proliferative signals (Liu and Levens, 2006). Heterodimerization of Myc with Max is obligatory for binding to the E-box sequence, CACGTG, leading to modest levels of transcriptional activation of genes proximal to Myc-Max binding sites. Such activation occurs through recruitment of multiple complexes that modify chromatin and/or stimulate RNA polymerase activity (for reviews see (Adhikary and Eilers, 2005; Amati et al., 2001; Cole and Nikiforov, 2006)). Moreover Myc can act to repress transcription by forming an inhibitory complex with Miz-1, a BTB-POZ domain activator (Adhikary et al., 2005; Staller et al., 2001) (for review see (Kleine-Kohlbrecher et al., 2006)).
A distinct group of bHLHZ proteins, the Mxd family (Mxd1-4 and Mnt, previously known as the Mad family), whose members also dimerize with Max and recognize E-box sites in DNA, act as antagonists of Myc function. Mxd proteins repress transcription through their association with the mSin3 co-repressor complex, which contains histone deacetylase (HDAC) activity (for reviews see (Hooker and Hurlin, 2006; Rottmann and Luscher, 2006)). Several lines of evidence indicate that Mxd downregulates genes that are normally activated by Myc and that the cellular proliferation and growth promoting activities induced by Myc are inhibited by Mxd overexpression (Amati and Land, 1994; Iritani et al., 2002; Roussel et al., 1996). These findings are consistent with the idea that the HDAC activity evinced upon Mxd-Max binding would reverse the HAT-induced histone acetylation resulting from Myc-Max binding. In general mxd gene expression is induced during terminal differentiation and cell cycle arrest, periods when Myc expression is normally downregulated, suggesting that Mxd proteins may initiate a silencing pathway for Myc target genes involved in cell proliferation and growth (Hooker and Hurlin, 2006; Rottmann and Luscher, 2006). This would imply that downregulation of Myc is not sufficient for target gene silencing. Indeed Mxd1 loss of function, especially in the context of p27Kip1 deletion, has been shown to impede differentiation of granulocytes and hematopoietic stem cells (McArthur et al., 2002; Walkley et al., 2005). However, not all Mxd family proteins have expression patterns related to growth arrest. The Mnt protein is expressed in quiescent and differentiating cells but is also readily detected, along with Myc, in actively proliferating cells (Hurlin et al., 1997). The simultaneous presence of Myc and Mnt is thought to reflect a balanced and dynamic regulation of histone acetylation and transcription at E-box binding sites.
The identification of dMyc, dMax and dMnt in Drosophila and the absence of any paralogs has greatly facilitated genetic analyses of these proteins and their functions (for recent reviews see (de la Cova and Johnston, 2006; Gallant, 2006)). Many crucial properties of the Max network have been conserved in flies, including heterodimerization of dMyc and dMnt with dMax, E-box recognition, transcription activation by dMyc-dMax, and Sin3 binding and repression by dMnt-dMax (Gallant et al., 1996; Loo et al., 2005). Furthermore dMyc can co-transform murine fibroblasts and rescue proliferation of c-Myc deficient mammalian cells while c-Myc can rescue lethal mutations of dMyc in Drosophila (Benassayag et al., 2005; Schreiber-Agus et al., 1997; Trumpp et al., 2001). An important conclusion from the Drosophila studies is that dMyc regulates cell and organismal size. Hypomorphic mutants of dm (diminutive, the gene encoding dMyc) are viable yet smaller and are comprised of smaller cells (Johnston et al., 1999) whereas a null mutation (dm4) leads to lethality due to arrested growth at an early larval stage, an effect closely linked to a dramatic failure in the growth of endoreplicating cells (Pierce et al., 2004). Mutation in the C-terminal bHLHZ region of dMyc also led to a profound decrease in the growth and endoreplication of germline and somatic cells in the ovary (Maines et al., 2004). By contrast tissue-specific overexpression of dMyc results in larger than normal cells in both mitotic and endoreplicating tissues, while widespread dMyc overexpression produces larger flies (Pierce et al., 2004) (Johnston et al., 1999). Analysis of clones in the wing disc shows that cells overexpressing dMyc increased in size at a faster rate than wild type cells while their division time was unaffected, indicating that dMyc predominantly influences cellular growth rate. The notion that Myc regulates cell growth is reinforced by results of many expression profiling studies showing that a significant fraction of genes whose transcription is altered by Myc in Drosophila and mammalian cells are involved in ribosome biogenesis, protein translation, and metabolism (Coller et al., 2000; Hulf et al., 2005; Li et al., 2005; O'Connell et al., 2003; Orian et al., 2003; Schlosser et al., 2005). Moreover Myc has been shown to stimulate transcription of ribosomal RNA encoding genes by direct binding to rDNA promoters in mammalian cells or by enhancing expression RNA polymerase I components in Drosophila (Arabi et al., 2005; Grandori et al., 2005; Grewal et al., 2005).
Taken together with evidence that c-Myc activates RNA polymerase III transcription of tRNAs and 5S ribosomal RNA, the studies described above indicate that Myc functions in both flies and vertebrates as a general transcriptional regulator of cell growth through stimulation of all three RNA polymerases. In this context it is interesting to consider the role of dMnt in growth control. Previous work has shown that overexpression of Mxd1 (Mad1) or dMnt attenuates rRNA transcription and results in smaller cells, suggesting an important regulatory role in growth (Iritani et al., 2002; Loo et al., 2005; Orian et al., 2005; Poortinga et al., 2004). However a null mutation in dmnt (dmnt1) produced a surprisingly mild phenotype. While the dmnt1 adult flies showed increased weight, larger cells, and decreased life-span compared to controls, they were viable and fertile, with no detectable developmental delays (Loo et al., 2005). This contrasts sharply with the lethal consequences of dm loss of function. In order to further explore the consequences of antagonism between dMyc and dMnt we have now examined the effects of dmnt mutation in a dm null background.
The dm4 dmnt1 line was generated by recombining dm4 (Pierce et al., 2004) and dmnt1 (Loo et al., 2005) X chromosomes and screening for recombinants by PCR. As controls we used precise excision lines isolated in the generation of dm4 or dmnt1.
For all experiments other than mitotic clone analysis, mutant and control X chromosomes were balanced with FM7i, Act-GFP and non-GFP mutant or control hemizygous males were analyzed.
For the mitotic clone experiments, ywnlsGFPFRT19A;70FLP70I-SceI/TM6B was constructed by recombining ywnlsGFPFRT18E (Davis et al., 1995) with FRT19A (Xu and Rubin, 1993) and crossing ywnlsGFPFRT19A to 70FLP70I-SceI /TM6B (Rong and Golic, 2000). dm4, dmnt1, and dm4dmnt1 were recombined with FRT19A.
Flies and larvae were grown at 25°C, unless otherwise noted.
For larval growth assays and analysis of larval tissues, eggs were collected onto grape juice agar plates and larvae of the appropriate genotype were transferred to fresh plates with yeast paste at approximately 24 hrs AED. For larval growth assays, the number of larvae at each stage was scored on the indicated days. When fewer than the original number of larvae was recovered, the remaining larvae were scored as “unaccounted”.
First instar larvae were fed 5-bromodeoxyuridine (BrdU) at 100µg/ml, dissected, and fixed in 70% ethanol/PBS. Imaginal discs were labeled with BrdU as described (Secombe 1998) and fixed in 60% ethanol/30% chloroform/10% acetic acid. Fixed and rehydrated tissues were hydrolyzed with 2 N HCl in PBS for 30 min and BrdU labeled cells were detected with mouse anti-BrdU (BD Biosciences; 1:100).
Antibody staining was carried out as described (Reis and Edgar, 2004). Rat anti-ELAV (1:10) and mouse anti-wg (1:100) were obtained from the Developmental Studies Hybridoma Bank (University of Iowa). Alexa568- and Alexa488-conjugated secondary antibodies (Molecular Probes) were used at 1:6000. Nuclei were labeled with 4’,6-Diamidine-2-phenylindole (DAPI) or propidium iodide. Alexa568-phalloidin (Molecular Probes) was used at 1: 500.
Mutant clones were generated using the FLP/FRT system (Xu and Rubin, 1993) in animals carrying a mutant FRT19 chromosome in trans to ywnlsGFPFRT19A, with 70FLP70I-SceI providing FLPase under heat shock control. Larvae were heat shocked 48 hr after egg deposition (AED) for 7 min in a 37°C water bath and dissected 120 hr AED.
Total RNA was isolated from larvae 24 hr AED using TRIzol Reagent (Invitrogen) and DNase treated with the DNA-free kit (Ambion). cDNA was generated by reverse transcription with SuperScript III Reverse Transcriptase (Invitrogen) and random hexamer primers (Invitrogen). Quantitative real-time PCR (qRT-PCR) analysis was performed in 20 µl reactions containing Power SYBR Green PCR Master Mix (ABI), 4 pmoles of each gene-specific primer and 0.012 (CG11837) or 0.0012 (all other genes) larval equivalents of cDNA. Cycling conditions in an ABI 7900HT instrument were 95°C for 10 sec, followed by 40 cycles of denaturing at 95°C for 15 sec and annealing and extension at 60°C for 1 min. Fold change relative to the expression of CamKII, which we have used previously as a control for gene expression in dm4 mutants (Pierce et al., 2004), was calculated using the 2−ΔΔCT method (Livak and Schmittgen, 2001). Eight biological replicates were analyzed for each primer set. The following primer pairs were used: pre-rRNA, GCTCCGCGGATAATAGGAAT and ATATTTGCCTGCCACCAAAA; RpI135, GGTTCGCCCCTGTACTGTTA and ATAGTGGCAGGTCGTGGAAC; RpI1, TGGCCTCATCAAACATCAAA and AAAAGGTTTGCCAACGTCAC; Tif1A, CTGTTGGCAAAACCTTGGAT and AATTGCACATGATGCGTGTT; Nop56, CAAGGGATTCACCGACAAGT and CAACTGATCCAGCAAAGCAA; fibrillarin, GGCGAGAAGATTGAGTACCG and GACACATGCGAGACTGTCGT; Nop60B, CTCTTGTGACCATGGTGTGG and GACATAACCGGTCAGCCACT; CG11837, ACAAGATGCCCAAGGTTACG and TCAGGATGTGCTGACCAAAG; dbe, CCGATTGCTACGTTTTGGTT and GGTGCACATTGTTCATGGTC; Rpl13A, ACGTGATCAAGAGCCTGGAG and CCTCGTAGCCGTAGGATTTG; RpS6, ACATGTCTGTGCTGGCTCTG and GTAGAGCTTGCGGATCTTGC; eIF6, AGGCTCAGTCAGGCCAGTAA and TCGTCGTTGTTCTCGAATTG; EF1beta, GTTCTTCTTTCGTCGGCAAC and GCGTTCAACTCCTTCAGTCC; CG6388, TTTCCATGGTCCAAGAGGAG and AGCAGACCGGAACTTTAGCA; CG12267, ATGCCTAGCCGAGAATGAAA and GCCTAAAGAAATCGCACTCG; Aats-tyr, CAGCGACTATCAGCTGTCCA and AGAGCCTGCAGACCAGGATA; Aats-ala, CCTGGTGAACACAGTGGTTG and ACCCAACTTTTCGATTGTGC; FK506-bp1, AGCGTTTCTCTGTTGGGCTA and CCGCTCTCATTCTCATCCTC; dMyc, CAGTTCCAGTTCGCAGTCAA and AGATAAACGCTGCTGGAGGA; dMnt, CACACAGGAGGTGCAACAAC and GGTGCAACTGATGATTGTGG; CamKII (control), GGACATGCACATACCCATCA and GCAGATGCACTTCGATGAAA.
Expression profiles were generated using a Drosophila spotted cDNA microarray manufactured at Fred Hutchinson Cancer Research Center based primarily on the first two Drosophila Gene Collection (DGC) clone set releases (Stapleton 2002). More details of the array and its content can be obtained at: http://www.ncbi.nlm.nih.gov/geo/index.cgi (accession number GPL1908).
Experimental and control total RNA samples (2 µg) were amplified using the Amino Allyl MessageAmp II aRNA Amplification Kit (Ambion, Inc., Austin, TX). Resulting aRNAs were subsequently coupled to Cy3 or Cy5 fluorophores (GE Healthcare Bio-Sciences Corp., Piscataway, NJ) using standard protocols (Fazzio 2001). Experimental and control labeled aRNA targets (4 µg each) were co-hybridized to microarrays for 16 hrs at 63°C and sequentially washed at room temperature in 1 × SSC and 0.03% SDS for 2 min, 1 × SSC for 2 min, 0.2 × SSC with agitation for 20 min, and 0.05 × SSC with agitation for 10 min. Arrays were immediately centrifuged until dry and scanned using a GenePix 4000B scanner (Molecular Devices Corporation, Sunnyvale, CA). Image analysis was performed using GenePix Pro 6.0.
For each array result, spot intensity signals were assessed for signal quality and those identified as poor quality were removed from further analysis. Spot-level ratios were background corrected, log2 transformed, and loess normalized using tools within the GeneTraffic MULTI software package (Stratagene, La Jolla, CA). Sample comparisons were performed using 3 biological independent experimental replicates that were each compared to a common reference. For each comparison, a dye-swapped technical replicate was also performed and the paired results were averaged and used as a single observation. Accordingly, a total of 6 arrays contributed to 3 independent observations. Differential expression analysis was performed using CyberT (Baldi 2001), a Bayesian t-statistic methodology that is designed for microarray analyses in studies with low replicate numbers.
For CyberT analysis, we employed the default window size of 101 and used a confidence value of 6. Differential gene expression was identified by ranking each gene’s corresponding Bayesian p-value and applying a false discovery rate correction of 5% (Benjamini 1995). A fold-change threshold of ± 1.8 was applied as an additional criterion. Accordingly, relative expression levels for a given gene with a p-value that satisfies the FDR condition and fold-change criteria were identified as differentially expressed. After additional filtering to remove results representing duplicate genes, data was obtained for 6370 features on the microarray.
To better understand the interaction between dMyc and dMnt, we generated a Drosophila line that is mutant for the genes encoding both dMyc (dm) and dMnt (dmnt). Null mutations in dm (Pierce et al., 2004) and dmnt (Loo et al., 2005), which are both located on the X chromosome, were recombined to yield dm4 dmnt1. We had previously shown that dm transcript and protein were undetectable at the time of hatching, 24hr after egg deposition (AED), in hemizygous male dm4 mutant larvae, which are the progeny of heterozygous mothers ((Pierce et al., 2004) and Fig S1). We also found that dmnt transcript was undetectable at 24hr AED, but that a low level of dMnt protein was present, in dmnt1 progeny of heterozygous mothers (Fig. S1). No RNA for either gene was detected in dm4dmnt1 double mutant larvae at 24 hr AED (Fig. S1).
When grown on standard Drosophila media, in the presence of heterozygous and wild type siblings, the dMyc null mutant dm4 has a severe growth defect. The majority of dm4 larvae die before or soon after the molt to the second instar and the remaining larvae arrest in the second instar (Pierce et al., 2004). In contrast, dm4dmnt1 larvae continue to develop and by five days AED are significantly larger than dm4 larvae (Fig 1A), although they are much smaller than control larvae. This phenotype was found to be identical for two lines isolated from independent recombination events. We were impressed by the level of growth rescue that resulted from removing the function of dMnt in addition to dMyc and wondered whether optimizing growth conditions would enhance rescue. We found that when the hemizygous mutant larvae were isolated from their heterozygous and wild type siblings and grown on yeast paste, the dm4 dmnt1 larvae continued to grow. In contrast, dm4 single mutants still reached their maximum size by five days AED on yeast. After an extended L3 larval stage, dm4dmnt1 mutant larvae reached a maximum size that was much larger than the maximum size of dm4 mutants, but still markedly smaller than control larvae at 5 days AED, immediately prior to pupariation (Fig 1B). In addition, a substantial fraction of dm4dmnt1 mutant larvae can continue to develop and form pupae (Fig 1C).
The extent to which survival and development were rescued in dm4dmnt1 mutant larvae is quantified in Fig. 1D. While 90% of dm4 larvae died before or during the second instar by six days AED, over 60% of dm4dmnt1 larvae remained alive at six days AED and approximately half of these had molted to the third instar (Figure 1D). Whereas dm4 larvae never formed pupae, approximately 35% of dm4 dmnt1 larvae pupariated. Although dm4dmnt1 larvae developed further than dm4 larvae, they grew more slowly than control larvae, taking approximately twice as long to pupariate, and the pupae that eventually formed were small and morphologically abnormal (Figure 1C and 1D). This partial rescue of the dm4 growth defect by the additional removal of dMnt function suggests that, in the context of the dMyc null mutant, dMnt negatively regulates growth. This negative regulation is relieved in the dm4dmnt1 double mutant, allowing the larvae to develop further. Moreover our data indicate that the growth of the double mutant is sensitive to the external environment.
Previous work has demonstrated that the growth defect in dm null animals is due, at least in part, to a defect in endoreplication in the polyploid larval tissues (Pierce et al., 2004). To determine whether the increased growth of dm4dmnt1 larvae relative to dm4 larvae is correlated with increased endoreplication, polyploid tissues from larvae at their maximum size (five days AED for dm4 and control larvae; twelve days AED for dm4dmnt1 larvae) were stained with DAPI to visualize DNA. The dm4 defects in salivary gland and fat body nuclear size and overall salivary gland size were significantly rescued in dm4dmnt1 mutants (Fig 2A), although dm4dmnt1 mutant nuclei and salivary glands were still significantly smaller than controls. Cell size was also significantly increased in dm4dmnt1 fat body relative to dm4 fat body, as indicated by staining with phalloidin, which binds to actin and visualizes cell borders (Fig 2A). These results indicate that by the time they reach their maximum size dm4dmnt1 larvae have undergone significantly more endoreplication than dm4 larvae, which correlates with their increased growth.
One possible explanation for the increased growth of endoreplicating tissues is that the rate or extent of endoreplication is increased. To address this possibility, we used bromodeoxyuridine (BrdU) to mark the nuclei of cells in which endoreplication had taken place. Larvae were fed BrdU from 30–50 hrs AED and endoreplicating tissues were fixed and stained to visualize BrdU incorporation. In anterior midgut and fat body, 95.2% (n=15, SEM=0.85) and 99.7% (n=10, SEM=0.23) of control cells, respectively, labeled with BrdU, indicating that they initiated at least one cycle of endoreplication during the labeling period. In contrast, as we had shown previously (Pierce et al., 2004), endoreplication in dm4 mutant midgut and fat body was significantly reduced, with 9.1% (n=25, SEM=1.30) and 14.5% (n=10, SEM=2.9) of nuclei incorporating BrdU, respectively (Figure 2B). Endoreplication was augmented in dm4dmnt1 larvae relative to dm4 larvae. BrdU incorporation was significantly increased (p<0.05, t-test) in midgut and fat body to 18.2% (n=16, SEM=1.73) and 23.3% (n=12, SEM=2.86) of nuclei, respectively. Taken together these results indicate that an increase in endoreplication, combined with the increased survival and duration of larval growth, leads to a significant rescue of the growth of polyploid larval tissues in dm4dmnt1 larvae.
Although the growth of dm4dmnt1 animals is significantly rescued relative to dm4 mutant larvae, we never observed eclosion or formation of pharate adults. In order for pupae to metamorphose, the imaginal discs, which are groups of diploid cells that will form the adult structures, must be present and correctly patterned. To determine whether dm4dmnt1 animals had imaginal discs that met these criteria, we dissected dm4dmnt1 3rd instar larvae approximately twelve days AED and compared them to wild type 3rd instar larvae that were dissected five days AED. Surprisingly, dm4dmnt1 larvae contained imaginal discs that appeared morphologically normal and were often nearly equivalent in size to control discs (Fig. 3). To broadly determine whether they were correctly patterned, we stained imaginal discs with antibodies directed against proteins involved in disc patterning. Eye-antenna discs were stained with anti-ELAV, which stains differentiated neurons. Staining of control discs showed a regular array of differentiated neurons in the posterior region of the eye disc (Fig. 3). Although dm4dmnt1 mutant eye-antenna discs were often somewhat smaller than control discs, staining with anti-ELAV indicated that the array of differentiated neurons was present and normal in appearance. Similarly, expression of Wingless (Wg), which is necessary for correct wing disc patterning was similar in control and dm4dmnt1 discs (Fig. 3). In addition, wing and leg discs everted within the first 24 hr after pupariation but did not develop further (data not shown). These data suggest that the dm4dmnt1 imaginal discs are morphologically normal and appear to grow disproportionately large compared with polyploid tissues (see below) suggesting that the failure of dm4dmnt1 animals to metamorphose into adults is not due to missing or incorrectly patterned imaginal discs.
To more closely examine the growth behavior of dm4dmnt1 imaginal discs, we isolated wing discs and salivary glands at the beginning and end of the third instar and compared their sizes. Control and dm4dmnt1 larvae were dissected within one hour of molting from the second to the third instar or at the end of the third instar. Nuclei were visualized by DAPI staining and tissues were photographed at 100X magnification for early third instar or 50X magnification for late third instar (Fig. 4A–D). At the beginning of the third instar both salivary glands and wing discs from dm4dmnt1 larvae are significantly smaller than those from control larvae. From early to late third instar, salivary glands of both genotypes approximately double in size, given that they look similar at 5X and 10X magnification. In contrast, dm4dmnt1 wing discs display a dramatically larger size relative to the salivary glands, than do control wing discs (Fig 3 A–D, compare with insets). This confirms that imaginal discs grow disproportionately large, relative to endoreplicating tissues, in dm4dmnt1 mutant larvae and indicates that the bulk of this growth occurs during the third larval instar.
Although dm4dmnt1 ultimately reach a size close to that of control discs, this occurs over a much longer time period. While control larvae spend approximately two days in the third instar, dm4dmnt1 spend five to ten days in this stage (Fig. 1). This suggests that, just as there is a defect in endoreplication in polyploid tissues, there may be a proliferation defect in the mitotic cells. To examine this, control and dm4dmnt1 imaginal discs and brains were labeled with BrdU to mark cells in S phase. Dissected larval tissues were incubated with BrdU for one hour prior to fixing and staining for BrdU incorporation. The pattern of S phases in dm4dmnt1 and control eye discs was very similar (Figure 4E, F, I, J). The band of synchronous S phases just posterior to the morphogenetic furrow, known as the second mitotic wave, was similar to controls in terms of the number of S phase cells. Likewise, the pattern of S phases in the brain was similar in dm4dmnt1 mutant and control brains, and dm4dmnt1 mutant brains were often similar in size to control brains (data not shown). In all cases the intensity of staining was similar, suggesting that the amount of BrdU incorporated per cell was not dramatically different between dm4dmnt1 mutant and control cells. No increase in cell death was detected in dm4dmnt1 wing discs by staining with an antibody to cleaved caspase-3 (data not shown). We conclude from these results that late in the third instar, dm4dmnt1 mutant imaginal disc cells are actively proliferating. The labeling of a similar proportion of cells at a similar intensity between dm4dmnt1 mutant and control cells suggests the ability of the mutant cells to enter the cycle and traverse S phase in response to developmental cues is not dramatically altered.
Although our BrdU labeling study suggests that proliferation is relatively normal in dm4dmnt1 mutant imaginal discs and brains, we may have failed to detect small differences in proliferation rate that could result in large cumulative differences in growth. To more directly assess whether dm4dmnt1 mutant mitotic cells grow more slowly than control cells, we used the FLP/FRT system (Xu and Rubin, 1993) to generate clones of cells that were homozygous for dm4 and dmnt1 single mutants, or dm4dmnt1. Mitotic recombination results in the formation of homozygous mutant (−/−) and sibling (“twin spot”) homozygous wild-type (+/+) clones of the same age in discs that are heterozygous for the mutation. Clone area can therefore be used as a measure of growth (Neufeld et al., 1998), and was assessed after clones had grown for 72 hr. Consistent with the mild phenotype of this mutation in the whole animal (Loo et al., 2005), dmnt1 −/− clones were similar in size to their +/+ twin spots (Fig. 5A and B). In contrast, dm4 and dm4dmnt1 −/− clones were only 6.5% and 8.8% as large, respectively, as their +/+ twin spots. Preliminary data indicate that clones derived from both mutants had approximately 10% as many cells as their wild-type twin spots. In addition, out of the 27 dm4 and 34 dm4dmnt1 +/+ clones examined 25.9% and 26.4% respectively lacked an associated mutant clone. The small size of the dm4 and dm4dmnt1 −/− clones relative to their twin spots indicates that the mutant cells grow and/or proliferate more slowly than control cells. This combined with the loss of approximately 25% of the mutant clones suggests that the mutant clones are also subject to the process of cell competition, in which slowly growing cells in a background of normally growing wing disc cells are eliminated. The juxtaposition of cells with differing levels of dMyc has been shown previously to lead to cell competition and the elimination of cells with lower dMyc levels (de la Cova et al., 2004; Moreno and Basler, 2004). Our data indicate that the sensitivity of dm4 mutant cells to cell competition is not alleviated by further loss of dMnt. This suggests that, unlike in endoreplicating cells, the dm4 growth defect in wing disc cells is not rescued by removing dMnt function. When the wing disc is comprised of only mutant cells, competition would not be a factor and the modest rescue of proliferation coupled with the temporal delay in development is likely to account for the disc rescue.
Because dMyc and dMnt are transcription factors, we expected that the growth effects of removing dMyc and dMnt would reflect transcriptional changes in the larvae. It is known from overexpression studies that the bulk of dMyc-responsive genes in Drosophila larvae are involved in growth, particularly in ribosome biogenesis and protein synthesis (Grewal et al., 2005; Orian et al., 2003). We used microarray technology to take a broad and unbiased approach to identifying important transcriptional changes in the dMyc and dMnt mutants (see Methods for details). RNA for analysis was isolated from larvae at 24 hr AED, just after hatching, a time at which all mutant and control larvae are similar in size and at the same developmental stage. Although larvae of the different genotypes are phenotypically indistinguishable at this time, we predicted that transcriptional differences would anticipate the phenotypic differences observed in dm4dmnt1 mutant animals. Additionally, by isolating RNA at 24 hr AED, we avoided the potential problem of distinguishing transcriptional differences due to genotype from those due to developmental staging, as the larvae of different genotypes diverge significantly with respect to developmental timing (see Fig. 1D). The dm4 and dmnt1 mutations were generated by the imprecise excision of P-elements in different genetic backgrounds. For each of these there is a precise excision line that can be used as a control. Because the dm4dmnt1 mutant line was generated by recombining dm4 and dmnt1, the resulting recombinant chromosome contains portions of each of the starting chromosomes. We found the two precise excision lines to be phenotypically indistinguishable.
Fluorescently labeled sample pairs were co-hybridized to a Drosophila cDNA spotted array containing ~12,000 features. When dm4 was compared to its precise excision control, 729 genes were identified as differentially expressed (see Materials and Methods). Of these, 552 genes were downregulated and 177 genes were upregulated (Supplementary Table 1). Among the genes down-regulated in the absence of dMyc, 65% overlapped with genes that were previously reported to be up-regulated in larvae in response to dMyc overexpression (only genes common to the arrays in both studies were examined) (Orian et al., 2003). In addition, the expression of 25% of the genes we found to be down-regulated in the absence of dMyc also decreased in response to dMyc inhibition by RNAi in cultured S2 cells, although the microarray platforms were different (Hulf et al., 2005). We used the FatiGO program (Al-Shahrour et al., 2004) to identify functional groups that were significantly overrepresented in the differentially expressed gene sets, relative to the remaining genes in the data set. Among the genes that were downregulated in dm4, and for which FatiGO obtained a functional annotation, all of the functional groups that FatiGO identified as significantly overrepresented were related to protein synthesis (Table 1; Supplementary Table 2).
Among the genes that were up-regulated in the absence of dMyc, only ten were identified by FatiGO as belonging to overrepresented functional groups. These were involved defense response and the response to pheromones (Supplementary Table 2). Activation of this group of genes may indicate that the animals are under stress, which would be consistent with their poor survival. Because dMyc is predominantly a transcriptional activator in flies and mammals, genes whose expression increases in the absence of dMyc are unlikely to be direct targets. For this reason, we did not analyze this group of genes further.
To explain the phenotypic growth rescue we next attempted to identify genes whose expression is decreased in dm4 and that are less affected or rescued in dm4dmt1, by direct microarray comparison of dm4 and dm4dmnt1 mutants. Although genes were identified as differentially expressed by this method (data not shown), many were not verifiable and this experiment failed to provide us with additional insight. Because the microarray analyses may not have been sufficiently sensitive to detect subtle expression differences we turned to a directed quantitative PCR approach.
The dm4 microarray results suggest that, consistent with other studies, loss of dMyc function primarily affects genes involved in protein synthesis. To verify these results and to determine how this class of genes is affected by the additional loss of dMnt function, we selected genes that were involved in several aspects of protein synthesis for quantitative real-time PCR analysis. These included rRNA synthesis and RNA Polymerase I (PolI) function, rRNA processing, ribosomal proteins, translation factors, and tRNA processing and RNA Polymerase III (PolIII) function. Within each category we chose genes with E boxes, which are more likely to be direct targets of dMyc and dMnt and those without E-boxes. These genes had been previously identified as dMyc regulated either by expression profiling or dMyc binding (Orian et al., 2003). We also included an example of an E box-containing gene that is not directly involved in protein synthesis. For the genes in this analysis, we verified that the relative gene expression difference in the dm4 precise excision as compared to the dmnt1 precise excision was no greater than 1.6-fold (average, 1.31; range, 1.03–1.59). The dmnt1 precise excision was used as a control in the experiments shown.
All of the dMyc responsive genes we tested were down-regulated by at least two-fold (−1 on a log2 scale) in dm4 mutant larvae (Fig. 6). While the expression of all of these genes was also decreased in dm4dmnt1 mutant larvae, a subset was significantly less severely affected in dm4dmnt1 mutant larvae, indicating a partial rescue of expression, relative to dm4 mutant larvae. Genes predicted to be involved in rRNA processing, including Nop56, fibrillarin, Nop60B, and CG11837, were the most significantly rescued (p<0.001). In addition, genes involved in rRNA synthesis, expression of pre-rRNA and RpI1 (PolI subunit) were also partially rescued (p<0.05). Expression of the ribosomal protein RpS6 and the translational initiation factor eIF6 was rescued to a similar extent. No other genes tested were significantly rescued. Interestingly, the presence of an E box was neither necessary nor sufficient for rescue of expression in dm4dmnt1 mutant larvae. Consistent with the increased cell growth observed in dmnt1 mutant animals, the expression of most dMyc responsive genes tested was up regulated in dmnt1 mutant larvae, although in all cases by less than two-fold. These results, together with the dm4 microarray results, suggest that defects in protein synthesis could largely account for the growth defect in dm4 mutant larvae and that an increase in ribosome biogenesis, resulting primarily from increased rRNA synthesis and processing, is likely to play a role in the growth rescue in dm4dmnt1 larvae.
Our experiments show that Drosophila larvae lacking both dMyc and dMnt display increased viability, grow significantly larger and advance considerably further in larval development than dMyc single mutants. We found that dm4dmnt1 mutant larvae reached a maximum size that was much larger than the maximum size of dm4 mutants, which are terminally growth arrested at an early larval stage. In addition, a substantial fraction of dm4dmnt1 mutant larvae can continue to develop and form pupae, although they fail to metamorphose (Fig 1). In the following sections we discuss the implications of these findings and speculate on a mechanism through which loss of dMnt function may rescue growth of dMyc null larvae.
The most dramatic defect in larvae lacking dMyc is the failure of endoreplicating cells to attain both wild type size and DNA levels (Pierce et al., 2004). Similar effects have been observed in germline and somatic follicle cells mutant for dm (Maines et al., 2004). Because endoreplicating cells comprise the bulk of the early larva, and are responsible for most of the 200-fold increase in mass that occurs between embryogenesis and pupariation (Church and Robertson, 1966), it is likely that the defect in endoreplicating tissues accounts to a large extent for the very limited larval growth and the inviability in dm4 animals. The additional removal of dMnt results in increased cell and nuclear growth, which is accompanied by a modest increase in the fraction of cells undergoing endoreplication in the early larva as well as increased survival. We surmise that this increase in endoreplication and growth, combined with the extended larval period, is likely to account for the overall growth rescue. These results indicate that the strong larval growth arrest phenotype in the dMyc single mutant (dm4) is due, in part, to the presence of dMnt, whose activity as a repressor would be unopposed by dMyc in dm4. As a result, when dMnt is removed in the dm4 background, a significant amount of growth can occur in absence of dMyc (see below). A recent study in mammalian cells and Xenopus extracts has demonstrated that c-Myc plays a regulatory role in initiation of DNA replication through direct association, together with Max, with the pre-replication complex (Dominguez-Sola et al., 2007). It is possible that dMyc could drive endoreplication directly, rather than as a secondary consequence of increased growth, as has been previously proposed (Pierce et al., 2004). Further experiments will be required to determine if dMyc is involved in DNA replication in Drosophila and, if so, whether dMnt plays an antagonistic role.
In Drosophila, dMyc has been demonstrated to drive the growth of mitotic as well as endoreplicating cells. dMyc overexpression results in dramatically larger wing disc cells and analysis of dMyc overexpressing clones has demonstrated that dMyc drives cell growth with little effect on cell division time (Johnston et al., 1999). Clones of mitotic cells with reduced dMyc activity are smaller than clones of wild type cells, indicating that the mutant cells grow more slowly (Johnston et al., 1999) (Maines et al., 2004). As expected, overexpression of dMnt results in small cells and cell clones while dMnt loss of function produces larger cells in the adult wing as well as heavier animals (Loo et al., 2005). In mammals, primary mouse embryo fibroblasts (MEFs) and other cell types lacking c-Myc are unable to grow and proliferate in response to mitogenic signals (de Alboran et al., 2001; Mateyak et al., 1997) (Trumpp et al., 2001; Walker et al., 2005). However MEFs lacking both c-Myc and Mnt display accelerated proliferation following selection for cells that have escaped apoptosis (Nilsson et al., 2004; Walker et al., 2005). While no reports to date have addressed loss of both Myc and Mnt function in tissues or whole organisms, the MEF studies are generally consistent with our finding that the imaginal discs, which consist of mitotically dividing cells, are rescued in larvae lacking dMyc and dMnt.
We believe it likely that the increased growth of the endoreplicating tissues plays a significant role in the growth rescue of the whole animal and the discs. In particular, the fat body is a critical tissue for mediating the response to nutritional signals and is required to promote growth and proliferation in mitotic cells and larvae in a cell non-autonomous fashion (Britton and Edgar, 1998; Colombani et al., 2003; Martin et al., 2000). However, if larvae are starved after growth begins, mitotic cells continue to proliferate after endoreplication stops (Britton and Edgar, 1998). Thus is it likely that the increased growth of the endoreplicating tissues, particularly the fat body, promote the growth of the entire animal to a point at which the imaginal discs can continue to proliferate despite the limited growth of the endoreplicating tissues. In animals that lack only dMyc, there may be sufficient growth of the endoreplicating tissues to initiate growth of the animal as a whole, but not enough to sustain that growth to a point at which the discs can proliferate more independently.
Given the strong rescue of overall imaginal disc growth we were surprised to find that clones of imaginal disc cells lacking dMyc alone or lacking both dMyc and dMnt are the same size and significantly smaller than their sibling wild type clones (Fig. 5). This indicates that these cells are at a proliferative disadvantage. Because such mutant clones are surrounded by cells with higher levels of dMyc their dramatically smaller clone size can be ascribed to cell competition, a process resulting in elimination of more slowly growing cell populations from the disc epithelium. It has been well established that cell clones with lower dMyc levels relative to the surrounding population are at a competitive disadvantage and die through activation of apoptotic pathways (Johnston et al., 1999) (de la Cova et al., 2004) (Moreno and Basler, 2004) (for review see (Secombe et al., 2004)). In this regard it is interesting that dMnt loss of function has no discernable effect on the size of dm4 clones in a heterozygous background (Fig. 5). One interpretation is that the growth rescue resulting from dMnt loss is simply not sufficient to avoid cell competition induced apoptosis. Alternatively it is possible that dMnt does not influence the pathway through which reduced dMyc levels stimulate apoptosis in cells targeted for elimination.
The signals required for the growth and maturation of multicellular organisms include extrinsic environmental cues, such as nutrition, and intrinsic developmental signals, which may have cell-autonomous and cell non-autonomous effects (for recent reviews see (Edgar, 2006; Mirth and Riddiford, 2007)). The major pathways that regulate growth in response to nutrition in Drosophila are the insulin receptor and TOR signaling pathways. The TOR pathway in particular is responsive to amino acids and, with input from the phosphoinositide 3-kinase (PI3K) branch of the insulin receptor pathway, regulates ribosome biogenesis and translation. (see (Edgar, 2006; Wullschleger et al., 2006) for reviews). In Drosophila dMyc and dMnt have been shown to influence cell and organismal growth primarily through effects on transcription (Bellosta et al., 2005; Hulf et al., 2005; Orian et al., 2005; Orian et al., 2003). Yet many Myc- and Mnt-regulated transcriptional targets in flies and vertebrates are themselves involved in regulation of the translational machinery. For example Myc directly binds to and stimulates RNA PolI transcription of rDNA in mammalian cells (Arabi et al., 2005; Grandori et al., 2005) and indirectly stimulates RNA Pol I transcription in Drosophila (Grewal et al., 2005; Orian et al., 2005) (see below). Interestingly, strong mutations in, or inhibition of, components of the insulin and TOR pathways result in larval growth arrest phenotypes that are similar to the effect of amino acid starvation and the phenotype of the dMyc null mutant, dm4 (Oldham et al., 2000) (Britton et al., 2002). A weaker TOR mutant resembles the dm4dmnt1 mutant, in that it grows larger and the wing discs overgrow at the expense of the endoreplicating tissues. Indeed we also noted that growth of the dm4dmnt1 mutant was sensitive to environmental conditions such as nutrient source and presence of competition (Fig. 1A and B). However, clear molecular connections between the insulin and Tor pathways and dMyc/dMnt function have yet to be established.
Although the dm4dmnt1 larvae display delayed pupariation (fig. 1), the pupae are all very uniform in size, suggesting that they have reached a threshold for growth or size. The transition from feeding and growth to pupariation and metamorphosis is normally triggered by induction of the steroid hormone ecdysone. This occurs after larvae reach a minimum viable weight at which they possess sufficient nutritional stores to survive metamorphosis (reviewed by (Mirth and Riddiford, 2007)). However our experiments indicate that treatment with ecdysone is unable to rescue the delay in pupariation (S.P., S.M. data not shown) suggesting that ecdysone is not limiting for dm4dmnt1 pupariation. Since the dm4dmnt1 pupae are abnormal, and never result in the formation of adult animals, their extended larval period and other modulatory factors in larvae may interact with ecdysone to trigger the initiation of metamorphosis without the larvae reaching the minimum weight required to sustain metamorphosis. Thus the failure of dm4dmnt1 larvae to survive metamorphosis could be explained by their failure to accumulate sufficient nutritional stores. In addition, there may be a specific requirement for dMyc during pupal development that cannot be rescued by removing dMnt function. This latter possibility would be consistent with earlier work showing that a subset of genomic binding sites for dMyc do not bind dMax or dMnt (Orian et al., 2005; Orian et al., 2003) as well as a more recent study indicating dMax independent functions of dMyc (D. Steiger and P. Gallant, manuscript submitted) – because dMax coordinates binding of both dMyc and dMnt to common E-box sites it is unlikely that genes activated by dMyc independent of dMax would be repressed by dMnt.
This is the first analysis of gene expression changes in animals that are null for dMyc and we find that the affected genes are primarily involved in protein synthesis. Although it was previously known that dMyc was required for pre-rRNA transcription and that PolI function is required for dMyc-induced cell growth, we have demonstrated that dMyc is also necessary for the expression of a wide range of genes involved in ribosome biogenesis and translation. This is consistent with what is known about Myc target genes in Drosophila and mammals, in which Myc proteins have been shown to regulate the transcription of genes from all three polymerases thus controlling both ribosome biogenesis and subsequent translation (O'Connell et al., 2003) (Gomez-Roman et al., 2003) (Orian et al., 2003) (Arabi et al., 2005; Grandori et al., 2005) (Grewal et al., 2005) (D. Steiger and P. Gallant, manuscript submitted). Because larvae are primarily made up of endoreplicating non-dividing cells, it is not surprising that we found particularly strong overlap between genes for which expression decreased in the absence of dMyc and those for which expression increased when dMyc was overexpressed in larvae (Grewal et al., 2005), and less overlap with dMyc direct binding sites (Orian et al., 2003) and dMyc-responsive genes identified in Drosophila tissue culture cell lines (Hulf et al., 2005). Note that some of the genes detected are not direct targets of dMyc but are likely to be induced as a secondary response to Myc.
Given the growth rescue seen when dMnt function is removed in the context of the dMyc mutant, it was particularly interesting to find that the expression of only a subset of dMyc-responsive genes was altered in the double mutants relative to the dMyc single mutant. There was a modest, but significant, increase in pre-rRNA, which can in part be explained by an increase in PolI components such as Rpl1. There was a more significant increase in the expression of genes that are known or predicted to be involved in rRNA processing (Fig. 6). Interestingly, c-Myc has been shown to promote rRNA processing in a human B-cell line (Holzel et al., 2005; Schlosser et al., 2003). We surmise that an increase in both rRNA synthesis and processing leads to more efficient ribosome biogenesis and, ultimately, protein synthesis. Although transcription of only one of four ribosomal protein genes that we tested by qPCR was significantly rescued (Fig. 6 and data not shown), the availability of ribosomal proteins may be increased through post-transcriptional mechanisms or existing ribosomal proteins may be utilized more efficiently due to increased levels of processed rRNA.
Although it has been shown that dMyc and dMnt are capable of acting antagonistically, in a manner similar to their mammalian counterparts, the lack of a dramatic phenotype in flies lacking dMnt might suggest that dMnt is not normally playing a major regulatory role at dMyc target genes (Loo et al., 2005). Nonetheless we do observe a modest increase in expression of growth related genes in dmnt1 larvae indicating that in wild type flies dMnt is likely functionally required to limit cell growth during normal development (Fig. 6). Our data showing that dMnt loss produces partial alleviation of the repression of a subset of growth related genes in dm4 larvae is consistent with the notion that dMnt is involved in downregulation of growth gene expression upon loss of dMyc. In the absence of both dMyc and dMnt, active repression by dMnt is relieved and gene transcription is likely to be dependent on the presence of other transcription factors that may be gene-specific and normally act to regulate expression of specific genes in collaboration with Myc and Mnt. It is also possible that other Drosophila bHLH class activators that do not normally interact with dMyc-dMnt regulated promoters may play a role in the rescue of growth related gene expression when both dMnt and dMyc are absent. We propose that it is this rescue of growth related gene expression that leads to increased viability and development.
A similar model has been proposed to explain the effects of deletion of another antagonistic pair of transcription factors dE2F1 and dE2F2 (Frolov et al., 2001). dE2F1 is a transcriptional activator that is essential for normal proliferation and larval growth. dE2F2 is a transcriptional repressor that can inhibit dE2F1-dependent transcription by binding to the same DNA binding sites. The growth defects in larvae lacking dE2F1 can largely be suppressed by also removing dE2F2, indicating that the dE2F1 mutant phenotype is largely due to unchecked repression by dE2F2. Interestingly the E2F proteins are also subunits of the Myb-MuvB/dREAM complex that additionally comprises Drosophila Myb, required for the selective amplification of the chorion genes, and the Mip130, Mip120, and Mip40 repressors of DNA replication (Georlette et al., 2007). Loss of function mutations of Dm-Myb are lethal, while single Mip mutants and the double Dm-Myb;Mip mutants are viable, indicating that the Mip proteins are primarily responsible for the lethality of the single Dm-Myb mutants (Beall et al., 2004) (Beall et al., 2007).
Our results suggest that in order to activate genes required for growth and development, dMyc must overcome the repressive effect of dMnt. It is therefore likely that dMnt serves to refine or limit the activity of dMyc. Although the weak phenotypes associated with the dMnt single mutant suggest that this role is minor, it is important to note that dMnt mutants as well as mutants in the C. elegans ortholog of dMnt (mdl1), have a reduced lifespan (Loo et al., 2005; Murphy et al., 2003), indicating that the modulation of dMyc activity by dMnt is necessary for wild type fitness.
Figure S1. Expression of dMyc and dMnt in 24 AED larvae. (A) Expression of dm and dmnt, relative to the control gene CaMKII, was determined by real-time PCR for larvae of the indicated genotypes. (B) Gut tissue from control larvae and dmnt1 mutant larvae from heterozygous (+/−) or homozygous (+/+) mothers was dissected at 24 hr AED and stained with anti-dMnt antibodies (upper panels) or DAPI (lower panels).
We thank Savraj Grewal, Bruce Edgar, Julie Secombe, and Lenora Loo for reagents, advice, and critical readings of the manuscript. We are grateful to Peter Gallant for information prior to publication and to Mike Botchan for encouraging us to generate the double mutants. We also thank Ilan Davis and the Bloomington Drosophila Stock Center for flies, Jimiane Ashe, Ryan Basom, Andy Marty for assistance with microarray experiments, and the Eisenman and Edgar labs for many helpful discussions. This work was supported by grant R37CA57138 from the NCI/NIH to R.N.E.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.