Imaginal discs dual control at G1 S and G2M

Cells in the developing imaginal discs, which form much of the adult body during metamorphosis, exhibit a cell cycle that incorporates both modes of control described above (Fig. 2C). Each imaginal disc has 10—50 cells in the newly hatched larva, and these proliferate to as many as 100 000 cells before differentiating into an adult structure such as a wing, leg, or eye (Bryant & Simpson 1984). Imaginal cells arrest in G1 during mid-embryogenesis and require the influx of nutrients from larval feeding to reactivate their cell cycle. Their first S phase occurs after a substantial (6x) increase in cell mass (Madhavan & Schneiderman 1977) suggesting that, as in the ERTs, cell growth may be a limiting parameter for progression of their cell cycle. The disc cell cycle averages 8 h and includes extended G1 and G2 periods (Graves & Schubiger 1982, Fain & Stevens 1982). During later disc development string is limiting for G2/M progression, whereas cyclin E expression limits G1/S progression (Milan et al 1996a,b, Neufeld et al 1998). Both phase transitions appear to be regulated, since disc cells typically spend several hours in each of G1 and G2, and either phase can be truncated by overexpressing the limiting regulator (Neufeld et al 1998).

We've used clonal activation or inactivation of genes, coupled with fluorescence-activated cell sorting (FACS) and quantitative in situ analysis, to probe the mechanisms of cell cycle control in imaginal wing and eye discs. Our initial tests showed that specific cell cycle regulators such as String, cyclin E, E2F, and the E2F co-repressor RBF play little role in regulating growth of the tissue (Neufeld et al 1998). When the cell cycle is either sped up or slowed down by manipulating these genes, cell growth continues at nearly normal rates. Consequently cells with a faster cycle are smaller, and cells with a slower cycle are larger. In contrast, mutations such as Minutes, which reduce growth by slowing protein synthesis, effect a slower cell cycle without changing cell size. This implies that cell growth has a dominant, regulatory effect on the cell cycle, and that the cell cycle control apparatus must employ mechanisms for sensing cell growth and adjusting cell division rates to match. We doubt that such mechanisms simply sense cell size, since cell size varies greatly during disc development (Madhavan & Schneiderman 1977, Neufeld et al 1998).

Potentially illuminating findings come from experiments in which we studied the behaviour of disc cell clones overexpressing genes that stimulate cell growth. The d-Myc transcription factor (Johnston et al 1999), the small GTPase Ras1 (Prober & Edgar 2000) and PI3K (Weinkove et al 1999) are all required for normal imaginal cell growth. When overexpressed each gene increases growth, as measured by increased clone sizes. Interestingly, increased growth effected by d-Myc, Ras, or PI3K does not accelerate the disc cell cycle. Instead, increased growth is manifest entirely as increased cell size (Fig. 3). In each case the large, fast growing cells generated have a truncated G1 phase. These observations suggested that increased cell growth is 'sensed' by cell cycle components that control G1/S transitions, but that G2/M controls are insensitive to growth stimulation. To test this notion we coexpressed the G2/M limiter String along with d-Myc (Johnston et al 1999) or activated Ras (RasV12; Prober & Edgar 2000). In both cases this

FIG. 3. PI3K, d-Myc and Ras accelerate cell growth, but not cell division. The cartoon depicts the behaviour of cell clones, induced at 0 h, that overexpress d-Myc, PI3K or Rasv12. These cells grow faster and make larger clones, but cell division is not accelerated, and thus the cells are enlarged. Each treatment shortens Gl, but appears to have little effect on rates of G2/M progression. Co-expression of the G2/M driver, String, with d-Myc or Rasvl2 reduces the cell size and accelerates cell division, but does not suppress growth of the clone as a whole. Cell autonomous activation of dpp signalling (with Tkv^°, an activated receptor) or mg signalling {with A rm&N, an activated intracellular transducer) also accelerates cell growth and division equally. String expression alone has little affect on growth or proliferative rate (Neufeld et al 1998).

FIG. 3. PI3K, d-Myc and Ras accelerate cell growth, but not cell division. The cartoon depicts the behaviour of cell clones, induced at 0 h, that overexpress d-Myc, PI3K or Rasv12. These cells grow faster and make larger clones, but cell division is not accelerated, and thus the cells are enlarged. Each treatment shortens Gl, but appears to have little effect on rates of G2/M progression. Co-expression of the G2/M driver, String, with d-Myc or Rasvl2 reduces the cell size and accelerates cell division, but does not suppress growth of the clone as a whole. Cell autonomous activation of dpp signalling (with Tkv^°, an activated receptor) or mg signalling {with A rm&N, an activated intracellular transducer) also accelerates cell growth and division equally. String expression alone has little affect on growth or proliferative rate (Neufeld et al 1998).

reduced the cell size back to approximately normal, truncated the G2 phase and allowed the cells to divide faster than normal (Fig. 3). Importantly, co-expression of Stg with these 'growth drivers' did not decrease the overall growth rate, as measured in cell clones. This corroborated our hypothesis that the primary effect of increasing d-Myc or Ras activity was stimulation of cell growth, and that truncation of Gl was a secondary response.

How might cellular growth drive Gl/S transitions? A mechanism in which an unstable, translationally regulated protein limits Gl/S transitions provides one possible explanation. Consistent with this, we found that ectopic RasVl2 or d-Myc increased levels of cyclin E, and that this effect was post-transcriptional (Prober &

Edgar 2000). The 5'-untranslated region of the Drosophila cjclinH mRNA contains several short open reading frames (uORFs) which could potentially modulate its translation. Polymenis & Schmidt (1997) proposed that a uORF in the yeast G1 cyclin, Cln3, reduces initiation of translation at the downstream Cln3 translation start site. As a result, more ribosomes are needed to achieve efficient translation of Cln3. Since ribosome abundance is coupled to the cellular growth rate, this uORF renders translation of Cln3 growth-sensitive. Our data suggest that a similar mechanism could regulate production of cyclin E in the developing wing. Cyclin E would thus act as a 'growth sensor' to couple cellular growth rates to G1/S progression (Fig. 2B,C).

One intriguing aspect of these findings is that G2/M progression was not affected by d-Myc- or Ras-driven growth. This suggests that the G2/M regulator, String, does not act as a 'growth sensor' in the cycles we studied. Various observations indicate that, in late stage discs, string instead becomes subject to dominant transcriptional control by patterning genes, much as described above for embryonic cell cycles. This presumably results from the onset of differential expression of transcription factors that specify cell fates, and also interact with the string control region. string expression in late stage wing discs is repressed at the dorsoventral compartment border in response to wingless (wg), which organizes pattern in this region (Johnston & Edgar 1998). Consistently, the cis-acting element responsive to wg signalling lies several kilobases away from the string promoter (fragment 6.4 in Fig. 1). string also displays vein/intervein patterning in later wing development, presumably in response to EGFR, Notch or decapentaplegic (dpp) signalling.

Interestingly, string's pattern sensing capability appears not to affect the cell cycle significantly during early disc growth. Cells in early discs cycle very rapidly, express string RNA constitutively and have a short G2 period (Fain & Stevens 1982, Neufeld et al 1998). This suggests that their cycle is primarily regulated at G1/S transitions and that, as in the ERTs, cell growth may be the limiting parameter. As development progresses however, distinct on/off patterns of string expression resolve and disc cells slow their cycle by accumulating in G2. In this way patterned control of G2/M by string appears to 'check' cell proliferation (and hence growth), even when growth promoters like PI3K or d-Myc are inappropriately activated. The contrast to the ERT cell cycle, which lacks this fail-safe check and runs wild in response to ectopic d-Myc, is intriguing.

How are G1/S and G2/M progression normally co-ordinated in the disc cell cycles? In closing it is pertinent to note that modulating pattern formation signals directly can indeed drive balanced cell growth. Specifically, cell autonomous activation of either dpp or wg signalling (two patterning factors required for normal wing growth) gives rise to cells that grow faster than normal, divide faster than normal, and retain a normal cell size (Fig. 3). Thus,

FIG. 4. Summary of imaginal disc cell cycles. Here we emphasize that patterning signals such as DPP and WG can accelerate both cell growth and cell division to the same degree ('balanced growth'). Hence these signals must stimulate both Gl/S and G2/M progression as well as cell metabolism. Coupling relationships observed in numerous experiments are shown.

FIG. 4. Summary of imaginal disc cell cycles. Here we emphasize that patterning signals such as DPP and WG can accelerate both cell growth and cell division to the same degree ('balanced growth'). Hence these signals must stimulate both Gl/S and G2/M progression as well as cell metabolism. Coupling relationships observed in numerous experiments are shown.

while dpp and wg signalling stimulate cell growth and G1/S progression, they must also augment the activity of the G2/M limiter string (Fig. 4). Determining the specific connections that link these patterning signals to cell growth and the cell cycle promises many interesting surprises.

0 0

Post a comment