The bantam Gene Regulates Drosophila Growth
David R. Hipfner, Katrin Weigmann, Stephen M. Cohen


We report here the consequences of mutations of a novel locus, named bantam, whose product is involved in the regulation of growth in Drosophila. bantam mutant animals are smaller than wild type, due to a reduction in cell number but not cell size, and do not have significant disruptions in patterning. Conversely, overexpression of the bantam product using the EP element EP(3)3622 causes overgrowth of wing and eye tissue. Overexpression in clones of cells results in an increased rate of cell proliferation and a matched increase in cellular growth rate, such that the resulting tissue is composed of more cells of a size comparable to wild type. These effects are strikingly similar to those associated with alterations in the activity of the cyclinD-cdk4 complex. However, epistasis and genetic interaction analyses indicate that bantam and cyclinD-cdk4 operate independently. Thus, the bantam locus represents a novel regulator of tissue growth.

MOST animals grow to a characteristic and reproducible size. Although the final size of the component parts of an animal can be greatly influenced by environmental factors such as nutrition, organ growth rates and size are also controlled by mechanisms intrinsic to the developing organs themselves (Bryant and Simpson 1984). The size of a given animal or organ is determined in large part by the number and size of its constituent cells. Consequently, the processes of cell division, cell death, and cell growth must be carefully regulated during development to ensure the correct and proportionate size of the adult animal (Conlon and Raff 1999).

The intrinsic and environmental mechanisms controlling growth have been the focus of considerable recent attention. Studies of the growth and development of Drosophila imaginal discs have begun to address the relative importance of cell growth and cell division in determining organ and body size (Edgar 1999; Lehner 1999; Oldhamet al. 2000a; Stocker and Hafen 2000). Imaginal discs are the larval structures from which all adult epidermal structures of the fly are derived. These epithelial sacs arise as small clusters of 20–50 cells during embryogenesis (Cohen 1993). In the span of ∼3 days during the larval instars, disc cells proliferate rapidly and increase in number ∼1000-fold. As in higher organisms, imaginal disc cell divisions are regulated at G1-S and G2-M transitions. In addition, disc cell divisions are thought to be growth dependent, meaning that the cells normally do not divide until they have grown to a certain critical size or mass (Edgar and Lehner 1996). Experimental manipulation of cell division rates without corresponding changes in cell growth rates can have significant effects on cell size (Johnstonet al. 1977; Weigmannet al. 1997; Neufeldet al. 1998). In Drosophila discs, accelerating the cell cycle by genetic means results in normal-sized discs with more and smaller cells. Conversely, slowing the cell cycle produces normal-sized discs with fewer and larger cells (Weigmannet al. 1997; Neufeldet al. 1998). The fact that final tissue size was unchanged in both cases, within limits, indicates that tissue growth is likely not regulated at the level of cell cycle control per se.

In contrast, a number of Drosophila genes have been identified that affect tissue growth directly. These genes all have in common the ability to regulate cellular growth rates. For example, genes encoding components of the insulin/phosphatidylinositol-3-kinase (PI3K) signaling pathway have been found to be instrumental in determining organ and body size. Overactivation of the pathway in imaginal discs causes tissue overgrowth by increasing the rate of cell growth. In the absence of a sufficient corresponding increase in the rate of cell division, this causes cells to divide at a larger than normal size (Goberdhanet al. 1999; Verduet al. 1999; Weinkoveet al. 1999; Gaoet al. 2000). Conversely, decreased pathway activity reduces tissue growth by producing smaller cells (Böhniet al. 1999; Goberdhanet al. 1999; Montagneet al. 1999; Verduet al. 1999; Weinkoveet al. 1999; Gaoet al. 2000), and in some cases also by reducing cell number (Böhniet al. 1999; Weinkoveet al. 1999). The tumor-suppressor genes TSC1 and TSC2 restrict tissue growth by regulating cell growth rates via the insulin/PI3K pathway (Gao and Pan 2001; Potteret al. 2001; Taponet al. 2001). In addition to components of the insulin/PI3K pathway, Drosophila homologs of ras, myc, and TOR have been shown to promote cell growth (Johnstonet al. 1999; Oldhamet al. 2000b; Prober and Edgar 2000; Zhanget al. 2000).

In all of the foregoing examples the effects on tissue growth rates were mediated primarily by regulating cell growth rates. Consequently, the normal balance between the rate of cell growth and the rate of cell cycle progression was lost. Cells grew too fast and divided at abnormally large sizes. In contrast, the complex composed of Drosophila cyclin D (cycD) and cyclin-dependent kinase (cdk) 4 controls tissue growth in a manner that keeps the rates of cell growth and cell cycle progression in balance (Dataret al. 2000; Meyeret al. 2000). Tissue overgrowth due to overactivation of cycD-cdk4 results from an increase in the number of normal-sized cells. Mutation of cdk4 reduced tissue size by reducing cell number rather than cell size. Thus cycD-cdk4 appears to control the rate of growth by coordinated regulation of cell growth rates and cell cycle progression rates. With the exception of ras, both these classes of growth genes (those affecting primarily cell growth vs. those affecting cell growth and division rates) appear to be primarily involved in growth regulation, as they have minimal effects upon tissue patterning.

To identify additional genes involved in regulating imaginal disc growth, we performed a gain-of-function genetic screen using the EP method developed by Rorth (1996). When combined with a source of GAL4, the EP element will direct expression of genomic sequences adjacent to its site of insertion. Previous studies have shown that a high proportion of EP elements direct GAL4-dependent overexpression of endogenous genes (Rorthet al. 1998). We restricted our analysis to genes involved in growth by screening for EP elements that showed GAL4-dependent effects on tissue size without disrupting pattern. Here, we report the identification of a locus that we call bantam (ban), which influences tissue growth rates. We present evidence that bantam is involved in coordinately regulating cell growth and cell division to regulate the rate of normal tissue growth.


Fly strains: The EP collection of 2300 lines (Rorthet al. 1998), as well as a new collection of 8500 independent strains carrying insertions of a modified EP element, termed EPg (Mataet al. 2000), were screened. The sevenless (sev), optomotor blind, MS1096, and engrailed (en) GAL4 drivers were used (Basleret al. 1989; Capdevila and Guerrero 1994; Fietzet al. 1995; Lecuitet al. 1996). EP lines were also screened for modifiers of the effects of overexpression of the tumor suppressor gene expanded (genotype: sevGAL4,UAS-expanded, svpAE127,ro1/+; + EP; Boedigheimer and Laughon 1993; Blaumueller and Mlodzik 2000). banL1170 is a P-element insertion allele, originally called l(3)L1170, and was obtained from the Bloomington Drosophila Stock Center. The armadil-loLacZ, FRT80B stock was provided by Jessica Treisman. Other mutant and transgenic strains are described in the following references: UAS-Dp110 (Leeverset al. 1996); UAS-EGFP (Denefet al. 2000); UAS-GFPNLS (Neufeldet al. 1998); HS-FLP1 (Struhl and Basler 1993); Actin5C > CD2 > GAL4 (Pignoni and Zipursky 1997); UAS-cycD,UAS-cdk4 (Dataret al. 2000); and cdk43 (Meyeret al. 2000).

Mapping and characterization of P-element insertions in the bantam locus: Insertion sites of EP(3)3622, EPg(3)30491, and EPg(3)35007 were determined by plasmid rescue according to standard procedures. Flanking sequences for banL1170, EP(3)3208, and EP(3)3219 were available from the Berkeley Fly Database. These P elements are inserted in chromosome 3L at cytological position 61C7-8. They are clustered within 12.3 kb of one another in an interval of 42 kb containing no known or predicted genes. EP(3)3622 contains two EP elements inserted in a back-to-back orientation at position 12,052 of genomic contig AE003469, with one basal promoter oriented proximally and one distally. The other P elements, with site of insertion in nucleotides relative to EP(3)3622 (+, further distal; –, further proximal), are EPg(3)35007 (–874); banL1170 (–173); EPg(3)30491 (–12); EP(3)3208 (+2040); and EP(3)3219 (+11,430). The banL1170, EP(3)3208, and EP(3)3219 chromosomes are homozygous lethal, but in each case the lethality can be attributed to another locus on the chromosome. banL1170, EP(3)3208, and EP(3)3219 are each viable in trans to the banΔ1 deletion. The revertant banL1170R1 chromosome, generated by excision of the banL1170 P element, lost the ban mutant phenotype but remained homozygous lethal.

Generation and molecular characterization of the banΔ1 allele: To generate mutants for the ban locus, P-element excisions of EP(3)3622 were generated. Excisions identified by the loss of the EP-element mini-w+ transgene were tested for complementation of the deletion Df(3L)Ar11, which removes from 61C3-4 to 61E. Excisions failing to complement this deficiency were analyzed by Southern blotting using genomic fragments derived from an EP(3)3622 plasmid rescue construct. One excision causing early pupal lethality, banΔ1, was found to delete sequences both proximal and distal to the original EP insertion site. The ends of the banΔ1 deletion were mapped by genomic PCR on DNA from homozygous mutant third instar larvae with primer pairs spaced at 5- to 10-kb intervals along the chromosome. Once approximate limits of the breakpoint were identified, a 2.9-kb PCR product spanning the junction was amplified from the same genomic DNA and sequenced.

Analysis of adult phenotypes: All crosses for size comparison were conducted under identical, uncrowded conditions. Crosses with enGAL4 for Figure 3, C and D, were carried out at 29°. All other crosses were at 25°. For the ban complementation analysis, males heterozygous for the P-element insertion being tested and the banΔ1 allele were crossed to banΔ1/TM2 females. In this way, each vial contained progeny of the tested genotype (e.g., P-element/banΔ1) and banΔ1/TM2 siblings. This allowed all measurements to be normalized relative to banΔ1/ TM2 sibling flies from within the same vial to eliminate the variability in adult body size resulting from differences in culture conditions between vials. Relative body mass was determined by weighing two or three sets of 20 male flies of each genotype from within a vial and taking the average. Final values in Table 1 are based on the average of at least two independent vials. Wing areas were measured using National Institutes of Health Image 1.59. To assess female fertility, virgin females (typically 40) of the appropriate genotype were crossed individually to wild-type males, and the number of viable adult offspring in each vial was counted ∼20 days later. Scanning electron microscopy was performed as described (Blaumueller and Mlodzik 2000).

Imaginal disc growth analyses: For all larval analyses, larvae were staged essentially as described (Neufeldet al. 1998). Embryos were collected for 3 hr. Twenty-four hours later, 65 newly hatched larvae of each genotype were transferred to fresh vials containing yeast paste. Discs from staged enGAL4, UAS-EGFP/+ larvae [with or without EP(3)3622] were dissected at 112 ± 1.5 hr after egg laying (AEL) and fixed in 4% formaldehyde. Discs were stained with 4′,6-diamidino-2-phenylindole (DAPI) to visualize the nuclei and analyzed by confocal microscopy. The posterior compartment and total disc areas were measured from the green fluorescent protein (GFP) and blue (DAPI) channels, respectively, using the histogram function of Adobe Photoshop.

Flow cytometry: Flip-out clones were induced at 72 ± 1.5 hr AEL in staged larvae of the genotype HS-FLP1/Act5C > CD2 > GAL4;UAS-GFP [with or without EP(3)3622 or UAS-Dp110] by heat shock at 38° for 1 hr. A total of 10–15 discs of each genotype were dissected in PBS at 112 ± 1.5 hr AEL, dissociated using trypsin, and stained with Hoechst 33342 as described (Neufeldet al. 1998). GFP content, cell cycle profiles, and forward scatter values were analyzed using a Cytomation MoFlo flow cytometer. Experiments were repeated three times with similar results.

Measurement of proliferation rates: Flip-out clones were induced at 72 ± 1.5 hr AEL in staged larvae of the genotype HS-FLP1/Act5C > CD2 > GAL4;UAS-GFPNLS [with or without EP(3)3622] by heat shock at 37° for 15 min. Discs were dissected at 112 ± 1.5 hr AEL and fixed in 4% formaldehyde. GFP-positive cells within each clone were counted by epifluorescence microscopy. Cell doubling times were calculated using the formula (log 2/log N)h, where N is the average number of cells/clone and h is the hours between heat shock and dissection (Neufeldet al. 1998). The experiment was repeated twice with nearly identical results.


bantam mutants produce small flies with fertility defects: In an overexpression screen for genes that affect tissue size, we identified several P-element insertions mapping within a 12.3-kb interval at cytological location 61C7-8 (Figure 1). P-element-mediated excision of EP(3) 3622 was used to produce a small deletion (Figure 1). The deletion is homozygous lethal at early pupal stages. Mutant larvae lack detectable imaginal discs. Flies heterozygous for the deletion and an independently isolated P-element insertion, l(3)L1170, are viable and normally patterned but are 15% smaller than sibling flies heterozygous for the deletion alone (Figure 2A, Table 1). The size reduction phenotype could be completely reverted by precise excision of the l(3)L1170 P element. Flies heterozygous for the deletion and a revertant of l(3)L1170 (called L1170 R1) are comparable in size to siblings heterozygous for the deletion alone (Table 1), indicating that the P-element insertion disrupts a gene required for normal growth of the fly. We therefore named the locus bantam, to indicate that the mutants are smaller than normal.

We examined in more detail the adult phenotypes resulting from decreased ban function by testing other P elements mapping to the deleted region for complementation of the banΔ1 excision allele. To eliminate the variability in adult body size resulting from differences in culture conditions between vials, measurements from flies heterozygous for banΔ1 and the P element were always normalized to siblings heterozygous for banΔ1 and the TM2 balancer chromosome reared in the same vial. EP(3)3622, EP(3)3208, EP(3)3219, EPg(3)30491, and EPg(3)35007 were each viable in combination with banΔ1. In addition to l(3)L1170 (which we renamed banL1170), EP(3)3622, EP(3)3208, EP(3)3219, and EPg(3)30491 caused a reduction in body size when in trans to banΔ1. In contrast, EPg(3)35007 had no effect (Table 1). Although smaller, the ban mutant flies were normally proportioned and the majority did not show significant patterning defects, suggesting that the product of the ban locus is primarily involved in regulating growth of all adult structures.

Another characteristic affected by the ban mutations was female fertility. All allelic combinations with banΔ1 that decreased adult size also caused a marked decrease in the average number of viable offspring produced by mutant females, whereas EPg(3)35007 did not (not shown). For example, almost all banL1170/banΔ1 females were sterile (2.5% fertile), in contrast to the 95% fertility rate of wild-type flies. As with the size reduction phenotype associated with banL1170, this effect on fertility could be completely reverted by precise excision of the P element, indicating that both phenotypes are due to the banL1170 insertion.

We also analyzed flies homozygous for the viable ban alleles EP(3)3622 and EPg(3)30491. In both cases, the homozygous flies showed the same reduced size and fertility defects as EP/banΔ1 trans-heterozygotes. Although the growth of EP(3)3622 and EPg(3)30491 homozygotes was affected to nearly the same extent as in the corresponding EP/banΔ1 flies (Table 1), the fertility of homozygous females of both genotypes was less severely reduced [50% of females sterile for EPg(3)30491/ EPg(3)30491 vs. 67.5% sterile for EPg(3)30491/banΔ1; 35% sterile for EP(3)3622/EP(3)3622 vs. 77.5% sterile for EP(3)3622/banΔ1; n = 40 in each case]. Neither EP(3)3622 nor EPg(3)30491 caused as great a reduction in adult size as banΔ1 when in trans to the other ban P-element insertions (not shown). We conclude that both insertions are likely hypomorphic ban alleles.

Overexpression of bantam causes overgrowth: Several of the ban EP-element insertions were identified in an overexpression screen. Four EP insertions produced noticeable overexpression phenotypes [EP(3)3622, EP(3) 3208, EPg(3)30491, and EPg(3)35007]. The more distally inserted EP(3)3219 element did not. When expressed under the control of enGAL4, EP(3)3622 increased the size of the posterior compartment in wing imaginal discs (Figure 3A). Measurement of the relative areas of the posterior and anterior compartments of discs from en-GAL4/+;EP(3)3622/+ larvae showed that a statistically significant increase of posterior to anterior area (P:A) ratio (P < 0.001; Figure 3B). To examine this phenotype in more detail we measured the effects of EP(3)3622, EP(3)3208, EPg(3)30491, and EPg(3)35007 on growth in the adult wing. Total wing area and the ratio of P:A compartment areas were measured. To exclude effects due to expression of enGAL4 in the region between veins 3 and 4 (Blair 1992), we used the area bounded by veins 1 and 3 as an estimate of anterior area and the area bounded by vein 4 and the posterior margin as an estimate of posterior area (illustrated in Figure 3D). enGAL4-driven expression of EPg(3)35007, EP(3)3208, EPg(3)30491, and EP(3)3622 caused statistically significant increases in the ratio of P:A areas compared to enGAL4/+ wings (P < 0.001; Figure 3C). EP(3)3622 had the strongest effect and was the only EP line to cause a statistically significant increase in the overall size of the wing (8%; P < 0.001; Figure 3C). Overgrowth of the posterior compartment occurred at the expense of the anterior compartment for the other EP lines, since there was no increase in overall wing size. This was also the case for EP(3)3622, because the magnitude of the increase in wing size was less than the relative increase in size of the posterior compartment. Only minor patterning abnormalities were observed in these wings (Figure 3D), suggesting that this EP element directs expression of a factor primarily involved in size regulation.

Figure 1.

—Map of the bantam locus (61C7-8). (A) Map showing the positions of genes surrounding the EP(3)3622 insertion site. Orientation and annotation numbers are indicated for several upstream and downstream genes. The region deleted in the banΔ1 allele is indicated below the map. (B) Magnification of the region deleted in the banΔ1 allele, showing the relative locations of P elements inserted in this interval. The orientation of the basal promoter is indicated by an arrow for each EP element. EPg(3)30491 and EP(3)3622 each contain two P elements inserted at the same position, oriented in opposite directions. EP elements producing a GAL4-dependent overexpression phenotype are indicated in italics. P elements producing a mutant phenotype in trans to banΔ1 are boxed.

The effects of EP(3)3622 overexpression are not limited to the wing. Expression of EP(3)3622 in cells behind the morphogenetic furrow using the gmrGAL4 driver caused bulging of the eye (Figure 3E), suggesting extensive overgrowth. The eyes were also externally rough. EP(3)3622 overexpression with appropriate drivers also caused duplication or triplication of interommatidial bristles in the eye and of macrochaete in the notum (data not shown). Similar effects on notum macrochaete have been described (Abdelilah-Seyfriedet al. 2000). These phenotypes are consistent with overproliferation of sensory organ precursor cells.

bantam mutant wings have fewer, but normal-sized, cells: A number of mutants affecting growth of adult flies have been identified. Viable mutants of Drosophila myc and certain components of the insulin/PI3K signaling pathway have been shown to produce small, normally patterned adult flies. The decreased size of these animals is due in large part to a reduction in the size of cells in the adult (Chenet al. 1996; Böhniet al. 1999; Johnstonet al. 1999; Montagneet al. 1999). Flies lacking the product of the cdk4 gene are also reduced in size. In this case, however, the size deficit is the result of a decrease in the number of adult cells, rather than of effects on cell size (Meyeret al. 2000). We analyzed wings from ban mutant flies to determine whether the growth deficit in these animals is due to decreased final adult cell size and/or cell number. Total wing areas were determined, and cell size was measured by number of wing hairs per unit area (each cell in the wing blade produces a single hair). Measurements for each allele in trans to banΔ1 were normalized to banΔ1/TM2 siblings. Wings from ban mutant flies were 9–13% smaller than banΔ1/TM2 siblings (Figure 2B, Table 1). For banL1170, EP(3)3622, and EPg(3)30491 over banΔ1, the decrease in wing size was not due to a decrease in cell size. The number of hairs per unit area was not significantly different in these mutants vs. the corresponding control wings (Table 1). Instead, the decreased wing size was entirely attributable to a reduction in the number of cells in mutant wings by up to 12%.

The EP(3)3219 insertion behaved differently, as the reduced wing size in EP(3)3219/banΔ1 was due primarily to a reduction in cell size rather than cell number. Interestingly, EP(3)3219 had little or no effect on growth when combined with other P-element alleles [EP(3)3219/banL1170, 103% of sibling banΔ1/TM2 body mass; EP(3)3219/EPg(3) 30491, 102%; EP(3)3219/EP(3)3622, 97%; averages from two independent vials]. These observations suggest that EP(3)3219 affects a genetically separable locus and that it is not an allele of ban.

Figure 2.

bantam mutations reduce body and wing growth. (A) Comparison showing the reduced body size of banL1170/ banΔ1 male and female flies compared to banΔ1/TM2 siblings. (B) Comparison of wing sizes from banL1170/banΔ1 and banΔ1/ TM2 sibling flies. In the overlay the banL1170/banΔ1 wing is shown in red and the banΔ1/TM2 wing in green.

The cellular effects of bantam overexpression are distinct from the insulin signaling pathway: To examine how overexpression of bantam causes tissue overgrowth, we examined cell size and cell cycle profile in clones of EP(3)3622-expressing cells. EP(3)3622-expressing clones marked by coexpression of GFP or control GFP-expressing clones were induced at the end of second instar and allowed to grow until late third instar. Coexpression of GFP was used to identify and sort the EP(3)3622-expressing cells, which were directly compared for cell cycle phasing and cell sizes with GFP-negative wild-type control cells from the same disc. There was no apparent difference in the distribution of EP(3)3622-expressing and wild-type cells in the G1, S, and G2 phases of the cell cycle (Figure 4A). Cell size was compared using forward scatter values (Neufeldet al. 1998). We observed a subtle difference between the EP-expressing and GFP-expressing control cells. Cells in control GFP-expressing clones were consistently slightly smaller than GFP-negative control cells (0.98; Figure 4B), whereas EP(3)3622-expressing, GFP-positive cells were consistently slightly larger (1.01; Figure 4C). Thus EP(3)3622-expressing cells were 3% larger than control GFP-expressing cells. Although reproducible, it is unclear whether this small difference is meaningful. For comparison, cells expressing the Dp110 catalytic subunit of PI3K, a known positive regulator of cellular growth rates, were 28% larger than control GFP-expressing cells, consistent with previous reports (Weinkoveet al. 1999; Figure 4D). These results indicate that EP(3)3622 expression has relatively little effect on cell size during the period of wing imaginal disc growth, consistent with the analysis of ban mutants.

These observations suggest that the mode of action of ban is distinct from Dp110 and other positive regulators of the insulin signaling pathway (Weinkoveet al. 1999). Although EP(3)3622 and Dp110 cause tissue overgrowth, they appear to have different effects at the cellular level. Activation of insulin signaling causes tissue growth by increasing the rate of cell growth more than the rate of cell division so that the overgrown tissue contains larger cells. EP(3)3622 causes tissue overgrowth, but does not cause a comparable net increase in cell size. This suggested that EP(3)3622-induced tissue overgrowth is coupled with an increase in the rate of cell division such that the overgrown tissue contains more cells, as has been demonstrated for the cycD/cdk4 complex (Dataret al. 2000). To test this we measured the cell-doubling rate by counting the average number of cells in clones allowed to grow for a defined time. Control GFP-expressing clones or EP(3)3622-expressing clones coexpressing GFP were induced in early third instar (72 hr AEL) and allowed to grow for 40 hr. The 187 control clones counted contained an average of 6.2 cells (median = 5), corresponding to a doubling time of 15.2 hr. In 213 clones expressing EP(3)3622, the average cell number was 7.0 (median = 7), corresponding to a doubling time of 14.2 hr. This difference is statistically significant (P < 0.005), and almost identical results were obtained in a second independent experiment. Together, these results indicate that EP(3)3622-overexpressing cells grow and divide more rapidly than control cells. Consistent with this observation, EP(3)3622 causes a large increase in the number of cells in regions of the adult wing in which it has been expressed (not shown). These observations suggest that ban coordinately regulates the rates of cell growth and cell division.

bantam does not interact genetically with cyclinD/ cdk4: The growth and fertility phenotypes associated with gain and loss of ban function are similar to those associated with alterations in the activity of the cycD-cdk4 complex (Dataret al. 2000; Meyeret al. 2000). Growth-impaired, viable mutants of both ban and cdk4 are composed of a smaller number of wild-type-sized cells. In both cases, female fertility is strongly impaired. The characteristics of tissue growth driven by cycD-cdk4 and ban are indistinguishable. In proliferating epithelial cells of the wing imaginal disc, both increase rates of cell cycle progression and growth in a coordinated manner, so that cell size remains normal. Acceleration of the cell cycle is apparently uniform, as no alterations in cell cycle phasing are detected. In postmitotic cells in the eye, cycD-cdk4 coexpression led to cellular hypertrophy. The resulting bulging, overgrown eyes are very similar in appearance to eyes in which ban overexpression was driven with the same GAL4 driver (compare Figure 3E with Figure 3G from Dataret al. 2000). These similarities raised the possibility that ban might act together with cycD-cdk4 to promote tissue growth. Consequently, we tested for genetic interactions between these growth regulators.

View this table:

Phenotypic analysis of bantam heteroallelic mutants

Figure 3.

—EPs inserted in the bantam locus cause GAL4-dependent tissue overgrowth in the wing and eye. (A) Imaginal discs from staged larvae expressing EGFP with (right) or without (left) coexpression of EP(3)3622 under the control of the enGAL4 driver were dissected at 112 hr AEL and fixed. Confocal images show discs with bright EGFP fluorescence marking the posterior compartment and less intense DAPI staining marking the entire disc (anterior is to the left). (B) Ratios of P:A compartment areas from enGAL4/+ control and enGAL4/ +;EP(3)3622/+ discs (n = 26 for each genotype; *, significantly different from control; P < 0.001). (C) Quantitation of the ratio of P:A compartment sizes and total wing area. EPs were expressed in the posterior compartment of the wing using the enGAL4 driver. Anterior and posterior areas were measured from eight female wings of each genotype. Shaded bars: graph of average P:A wing-area ratios for flies expressing EPg(3)35007, EP(3)3208, EPg(3)30491,or EP(3)3622 in the posterior compartment. Solid bars: graph of average total areas of the same set of wings. *, significantly different from enG4/+ controls (P < 0.001). (D) Comparison of enGAL4/+ control and enGAL4/+;EP(3)3622/+ wings. Anterior and posterior areas measured to calculate the P:A area ratios are outlined on the control wing. (E) Scanning electron microscopy images of heads from gmrGAL4/+ control and gmrGAL4/+;EP(3)3622/+ flies. EP(3)3622 expression in the eye causes roughening and bulging, suggesting extensive overgrowth.

Figure 4.

EP(3)3622 expression has little effect on disc cell size. Clones of cells expressing GFP alone or with either EP(3)3622 or wild-type Dp110 were induced at 72 hr AEL. Discs were dissected at 112 hr AEL and cells were dissociated and analyzed by flow cytometry. In A–D, data for GFP and EP/ transgene-expressing clonal cells and nonexpressing control cells from the same discs are in green and red, respectively. (A) Distribution of cells in G1 (2C DNA content), S (2C–4C), and G2 (4C) phases of the cell cycle for EP(3)3622-expressing and control cells. (B–D) Forward scatter analysis of cells from discs containing (B) control clones, (C) EP(3)3622-expressing clones, or (D) Dp110-expressing clones. Numbers represent the ratio of forward scatter values for GFP+/GFP– cells. Similar results were obtained in two additional independent experiments.

Importantly, both cycD and cdk4 are required to promote tissue growth. Expression of either alone is not sufficient (Dataret al. 2000). Therefore, to determine whether ban-driven overgrowth was dependent upon the activity of the cycD-cdk4 complex, we tested whether it could be blocked by removal of one of the components of this complex. We made use of the MS1096GAL4 driver, which directs GAL4 expression in the dorsal compartment of the wing disc early in larval development and more broadly throughout the developing wing pouch later (Milánet al. 1998). Expression of EP(3)3622 with MS1096GAL4 resulted in significant overgrowth of the entire wing. The effect was greater in the dorsal compartment, such that the wings curved downwards (not shown). MS1096GAL4/+;EP(3)3622/+ wings were 18% greater in area than control MS1096GAL4/+ wings (Figure 5A). Overexpression of ban in a cdk4 null mutant background (cdk43/cdk43) had no effect on the overgrowth phenotype. Wings of MS1096GAL4/+;cdk43/cdk43;EP(3) 3622/+ flies were 18% larger than those of MS1096GAL4/ +;cdk43/cdk43 flies (Figure 5A) and were also curved downward (not shown). As loss of cdk4 had no apparent effect on ban-driven growth, we conclude that ban does not promote growth by regulating the activity of the cycD-cdk4 complex.

One of the main functions of cycD-cdk4 is believed to be suppression of the function of “pocket” proteins, such as pRb. However, genetic analyses in Drosophila suggested that the effects of cycD-cdk4 in promoting cellular growth are mediated at least in part by unknown downstream targets, independent of the fly pRb homolog RBF (Dataret al. 2000). We tested whether ban might be such a downstream target. Overexpression of cycD-cdk4 with the enGAL4 driver resulted in overgrowth of the posterior compartment of the wing, significantly increasing the P:A ratio by 11% (Figure 5B). This overgrowth was unaffected by halving the gene dosage of ban (Figure 5B). enGAL4-driven cycD-cdk4 expression was also able to promote posterior compartment overgrowth to a comparable extent when the ban gene dosage was further reduced in the banL1170/banΔ1 allelic combination, although survival of these flies was poor (not shown). These observations suggest that the growth-promoting effects of cycD-cdk4 are not dependent upon ban levels.

To further evaluate the relationship between cycD-cdk4 and ban, we turned to an independent genetic assay for ban activity. Two of the ban EP insertions were identified initially in a genetic interaction screen as suppressors of the phenotype caused by overexpression of the expanded tumor suppressor gene. When misexpressed in the eye under the control of the sevGAL4 driver, expanded caused a reduction in eye size relative to wild type and external roughening and blistering (Figure 6, A and B; Blaumueller and Mlodzik 2000). Coexpression of EP(3)3622 almost completely suppressed this phenotype, restoring the eye to nearly wild-type size and appearance (Figure 6C). Reducing ban function had the opposite effect. Introducing one copy of the banΔ1 allele noticeably reduced the overall eye size and increased the blistering in the central and anterior regions of the eye (Figure 6D). In contrast, alterations in cycD-cdk4 activity did not alter the expanded overexpression phenotype. Coexpression of cycD-cdk4 with expanded increased the overall size of the eye, consistent with previous observations (Dataret al. 2000), but had little effect on the roughness and blistering (Figure 6E). Removing one copy of cdk4 had no effect (Figure 6F). The lack of a strong genetic interaction between expanded and cycD-cdk4 provides additional evidence that bantam is acting independently of this complex to promote coordinated cell growth and cell cycle progression.

Figure 5.

EP(3)3622 and cyclinD-cdk4 act independently to drive tissue growth. (A) Wing blade areas were measured from control flies and flies expressing EP(3)3622 under the control of the MS1096GAL4 driver in a wild-type or cdk43 mutant background. Seven female wings of each genotype were measured. Genotypes: MS1096GAL4/+; MS1096GAL4/+;EP(3)3622/+; MS1096GAL4/+;cdk43/cdk43; and MS1096GAL4/+;cdk43/cdk43;EP(3) 3622/+. *, significantly different from MS1096GAL4/+ (P < 0.001). **, significantly different from MS1096GAL4/+;cdk43/ cdk43 (P < 0.001). (B) CycD and cdk4 were coexpressed in the posterior compartment of the wing using the enGAL4 driver in a wild-type or banΔ1/+ background. Anterior and posterior areas were measured from six female wings of each genotype and expressed as P:A ratios. Genotypes: enGAL4/+; enGAL4/+;UAS-cycD,UAS-cdk4/+; and enGAL4/+;UAS-cycD,UAS-cdk4/banΔ1. *, significantly different from enGAL4/+ (P < 0.001).

Figure 6.

bantam, but not cycD-cdk4, interacts genetically with the expanded tumor suppressor. Scanning electron microscopy images of eyes from female flies raised at 25°. Genotypes: (A) sevGAL4/+. (B) sevGAL4,UAS-expanded, svpAE127,ro1/+. (C) sevGAL4,UAS-expanded,svpAE127,ro1/ EP(3)3622. (D) sevGAL4,UAS-expanded,svpAE127,ro1/ banΔ1. (E) sevGAL4,UAS-expanded,svpAE127,ro1/UAS-cycD, UAScdk4. (F) cdk43/+;sevGAL4,UAS-expanded, svpAE127,ro1/+.

Molecular characterization of the bantam locus: The banΔ1 deletion removes 21,147 nucleotides, extending from 5792 nucleotides proximal to 15,355 nucleotides distal to the EP(3)3622 insertion site, and fails to complement the deficiency Df(3L)Ar11 that removes from 61C3-4 to 61E. The banΔ1 deletion does not extend into the coding sequences of either of the two identified neighboring genes, CG3200 (Reg-2) or the predicted CG12030 (Figure 1). It is possible that noncoding or alternate exons of these genes might be located close to the EP elements. However, expression levels of CG12030, Reg-2, and CG13893 (see Figure 1) were comparable in banΔ1 mutant and wild-type third instar larvae as assessed by Northern blot analysis (not shown). This suggests that the mutant phenotypes associated with banΔ1 are not due to effects on expression of any of these nearby genes. Similarly, we did not detect upregulation of CG12030, Reg-2, or CG13893 or of the more distant genes CG17181, CG12189, or CG12015 by in situ hybridization analysis of wing discs in which EP(3)3622 was expressed.

It is possible that the ban locus may correspond to a gene contained at least in part within the ∼41-kb interval between CG12030 and Reg-2, which was not predicted in the CELERA/BDGP annotation. Although this intergenic region contains numerous short open reading frames, we have not found any meaningful homologies by BLAST sequence analysis. We have cloned two overexpressed transcription units mapping to this region that were first identified by in situ hybridization and Northern blot analysis of overexpression RNA with flanking genomic sequences as probes, or by RT-PCR using RNA from overexpressing larvae and an EP-element specific primer (Rorth 1996). However, neither of these transcripts is able to reproduce the overgrowth phenotype when expressed from a transgene (not shown). A more complete understanding of the mechanism of ban action will await molecular characterization of the gene product.


bantam is required for normal tissue growth: ban gene function appears to be important for regulation of tissue growth rates. Several EP elements inserted in this locus, most notably EP(3)3622, are capable of promoting substantial tissue overgrowth in the eye and wing in a GAL4-dependent manner. Conversely, ban mutations decrease tissue growth. Mutant phenotypes range from decreased body size to lethality. The strongest available allele is a small deletion that does not remove any known genes. This allele is pupal lethal and causes the absence of detectable imaginal discs. The simplest explanation for the reciprocal nature of gain-of-function and loss-of-function phenotypes is that EP(3)3622 is driving expression of the same transcription unit that is affected by ban mutations. This is further supported by the specific and reciprocal nature of the genetic interaction of gain and loss of ban function with the expanded tumor suppressor gene in the eye. However, this remains to be confirmed by molecular characterization of the locus. Growth regulation appears to be a primary function of ban, as EP(3)3622 expression does not cause significant patterning alterations, and ban mutant flies, although small, are proportioned normally.

bantam coordinates cell growth and division rates to regulate tissue growth: Our results suggest that ban regulates tissue growth by a mechanism that involves coordinated stimulation of cell growth and cell division. ban alters tissue growth through effects on cell number rather than cell size. Decreased ban function causes a reduction in cell number in the adult wing, but the surviving cells are of wild-type size, suggesting a coordinated decrease in the rate of cell growth and division. Activation of EP(3)3622 has the opposite effect on cell number, causing an increase in the rate at which imaginal disc cells proliferate. Despite this increased proliferation rate, cell sizes are little changed. These observations suggest that the rate of increase in cell division is coordinated with the rate of increase of cell mass when ban is overexpressed. The effects of ban on growth and fertility are remarkably similar to those of cycD-cdk4. However, we have found no evidence of a direct connection between cycD-cdk4 and ban. It seems unlikely that ban regulates growth by controlling the activity of cycD-cdk4, because ban-driven overgrowth is unaffected in the absence of cdk4. Similarly, cycD-cdk4-driven growth is unaffected by reduction of ban, indicating that ban is unlikely to be a downstream effector. We favor the view that ban and cycD-cdk4 act independently. The similarity in their growth phenotypes suggests that they may have some targets in common. However, as attested to by the differences in their interactions with expanded, they clearly can act differently as well.

Growth and pattern formation: The imaginal discs are patterned while they grow. The secreted signaling proteins Decapentaplegic (Dpp) and Wingless pattern the wing and leg discs along their main axes. Dpp and Wingless signaling are also required in some way for disc growth. The parts of the discs that produce the appendages are very small in flies lacking either signal (Spenceret al. 1982; Diaz-Benjumeaet al. 1994; Zeccaet al. 1995; Neumann and Cohen 1996). Cells unable to transduce the Dpp or Wingless signals display cell-autonomous defects in proliferation and are lost from the disc (Peiferet al. 1991; Burke and Basler 1996). To date it has not been reported whether loss of cells under these conditions is due to reduced proliferation or to reduced survival. However, recent studies suggest that Dpp signaling may directly influence cell proliferation in the wing disc (Martin-Castellanos and Edgar 2002). Wingless signaling has been shown in one situation to repress growth at late stages of wing development, in part by negative regulation of dmyc expression (Johnston and Edgar 1998; Johnstonet al. 1999). If Dpp and Wingless act directly to regulate tissue growth, we would expect them to coordinately regulate cell growth and cell division rates. It will be of interest to learn whether ban and/or cycD-cdk4 mediate the growth effects of these signaling molecules.

Compartments and imaginal discs as units of size control: Altering cell division rates does not alter compartment size, but can increase or decrease the number of cells per compartment (Weigmannet al. 1997; Neufeldet al. 1998). This is consistent with the effects of Minute mutations that vary the proportion of a compartment that can be contributed by the progeny of a single cell, without affecting compartment size or shape (Morata and Ripoll 1975). However, as first shown by Leevers et al. (1996), it is possible to alter the size of one compartment relative to another by manipulating activity of the insulin/PI3K pathway (Leeverset al. 1996). PI3K-induced overgrowth requires that the pathway be activated in all cells of the compartment. Clones of overgrowing cells do not affect the size of the compartment (Teleman and Cohen 2000). Thus a mechanism must exist that allows a population of cells to measure the size of the compartment. Interestingly, it has been found that altering the size of the compartment feeds back by an unknown mechanism to alter the shape of the Dpp morphogen gradient (Teleman and Cohen 2000).

Overexpression of ban with enGAL4 promoted significant overgrowth of the posterior compartment. We noted that posterior compartment overgrowth was compensated for by a nonautonomous reduction in the final size of the anterior compartment in most cases. This compensation suggests that total disc size may also be regulated to some extent during development. Only in the case of the strongest EP element, EP(3)3622, were total disc and wing size increased. These observations suggest that there may be multiple layers of size control operating during imaginal disc development. Morphogen gradients influence tissue growth. Tissue growth rates influence compartment size and morphogen gradient shape. Finally, size compensation mechanisms exist to control both compartment and disc size. At present, little is known about the size-sensing mechanisms, except that we can override them by stimulating cell and tissue growth rates by various experimental means. Identifying how size is measured during tissue growth poses a significant challenge.


We thank Ann Atzberger for flow cytometric analyses, Ann Mari Voie for preparing transgenic fly strains, and Christine Blaumueller for fly stocks and advice; Enrique Martin-Blanco for sharing information on the EP(3)3622 insertions prior to publication; and Pernille Rørth and members of the lab for helpful discussions. K.W. and D.R.H. were fellows of the European Molecular Biology Organization. D.R.H. was a fellow of the Human Frontiers Science Program Organization.


  • Communicating editor: T. Schüpbach

  • Received February 12, 2002.
  • Accepted May 9, 2002.


View Abstract