ABSTRACT
The Drosophila Ras1 gene is required for proper cell fate specification throughout development, and the loss-of-function phenotype of Ras1 suggests an additional role in cell proliferation or survival. A direct role for RAS1 in promoting cell proliferation, however, has not been established. We show that expression of an activated form of RAS1 (RAS1V12) during Drosophila imaginal disc development is sufficient to drive ectopic cell proliferation and hyperplastic tissue growth. In addition, expression of RAS1V12 induces widespread cell death in the imaginal discs, including cells not expressing the transgene, which results in ablation of adult structures. Loss-of-function mutations in the genes encoding RAF, MEK, MAPK and KSR dominantly suppress RAS1V12-induced cell proliferation. Furthermore, two RAS effector loop mutations (E37G and Y40C) that block the RAS-RAF interaction, also suppress RAS1V12-induced proliferation, consistent with a requirement for the MAPK cascade during the RAS1 mitogenic response. These two RAS effector loop mutants, however, retain some activity and can act synergistically with a MAPK gain-of-function mutation, suggesting that RAS1 may also act through signaling pathway(s) distinct from the MAPK cascade.
INTRODUCTION
Ras has been implicated in controlling both cell proliferation and differentiation, and its mutational activation is associated with a significant fraction of human tumors (reviewed in Lowy and Willumsen, 1993; Moodie and Wolfman, 1994). RAS activity is tightly controlled by a highly conserved signaling pathway, triggered by growth factor activation of Receptor Tyrosine Kinases (RTK; reviewed in Schlessinger, 1993; van der Geer et al., 1994). In turn, RAS activation leads to the sequential activation of RAF, MEK and MAPK, the highly conserved MAP Kinase cascade (reviewed in Daum et al., 1994).
Our understanding of the role Ras plays during cell differentiation has expanded largely through experiments in genetic systems such as the developing eye of Drosophila and vulva of C. elegans (review in Wassarman and Therrien, 1997; Sundaram and Han, 1996). Drosophila, for example, possess a single homologue of mammalian N-ras, H-ras and K-ras genes called Ras1 (Simon et al., 1991), which is activated in response to a variety of RTKs throughout development. Among other things, Ras1 is required for proper differentiation of the wing veins and the R7 photoreceptor cells (Diaz-Benjumea and Hafen, 1994; also see reviews Lu et al., 1993; Wassarman and Therrien, 1997). Expression of mutationally activated Ras1 (RAS1V12) under control of the sevenless enhancer/promoter (sev-Ras1V12) bypasses the normal requirement for Sevenless RTK activity and triggers R7 differentiation, and transforms the non-neuronal cone cells into ectopic R7 cells (Fortini et al., 1992). The R7 and cone cell precursors presumably are postmitotic since no mitogenic response to Ras1 signaling is observed. The sev-Ras1V12 transgene produces a rough eye phenotype, which has been used to isolate mutations in genes that act downstream of Ras1 during R7 differentiation (Karim et al., 1996).
We have a much more limited understanding of how Ras controls cell proliferation and how this differs from its role in differentiation. In the case of mammalian PC12 cells, for example, stimulation with NGF causes a sustained activation of RAS and MAPK and triggers differentiation, whereas stimulation with EGF causes a short-lived activation of RAS and MAPK and triggers proliferation (reviewed in Marshall, 1995). It is unclear, however, whether this phenomenon is specific to PC12 cells or reflects a more general mechanism. Interestingly, Serrano et al. (1997) showed that prolonged expression of mutationally activated H-ras (H-rasV12) induces senescence in human and rodent primary cell lines, whereas other groups have shown that expression of H-rasV12 can transform some immortalized mammalian cell lines but not others (reviewed in Lowy and Willumsen, 1993; Moodie and Wolfman, 1994). The varied results obtained using different cell lines illustrates the need for in vivo model systems for studying Ras-dependent cell proliferation, to better understand how Ras triggers different responses in differing cellular and developmental contexts.
The role of Ras during proliferation in Drosophila and C. elegans has been even less clear. Recently, Yochem et al. (1997) showed that, in C. elegans, let-60/Ras is not required for cell proliferation or cell survival. Further, let-60/Ras plays no apparent role in promoting cell proliferation, since an activating mutation in the endogenous gene produces only differentiation defects (reviewed in Sundaram and Han, 1996). By contrast, strong loss-of-function mutations in Drosophila Ras1, Ksr, Raf, MEK and MAPK are homozygous lethal and die as third instar larvae completely lacking imaginal discs, which constitute the bulk of the proliferating tissues during larval development (reviewed in Wassarman and Therrien, 1997). Similarly, homozygous mutant clones of Ras1, Ksr, Raf, MEK or MAPK in an otherwise heterozygous animal fail to survive to the adult stage, which suggests that RAS1 and the MAPK cascade are required for cell proliferation and/or cell survival (Diaz-Benjumea and Hafen, 1994; reviewed in Wassarman and Therrien, 1997). As yet a direct role for Ras1 signaling in promoting cell proliferation has not been established.
Drosophila imaginal discs, which are groups of undifferentiated monolayer epithelial cells that give rise to most of the adult structures, are ideal for studying cell proliferation control (reviewed in Edgar and Lehner, 1996; Bryant, 1996). During larval development imaginal discs proliferate exponentially until just prior to metamorphosis when they reach a stereotypical final size and shape (reviewed in Bryant, 1996). As seen during vertebrate development, imaginal disc cells do not have determined lineages but rather display variable proliferation patterns. Early in imaginal disc development each disc is subdivided into lineage compartments (anterior and posterior) that impose some restriction on the pattern of proliferation and act as units of growth control. Within each compartment, however, the proliferation patterns are heterogeneous and show no lineage dependence. Finally, little cell death is observed in the wing disc implying that disc size is achieved primarily by regulating the rate and pattern of proliferation (review in Edgar and Lehner, 1996; Bryant, 1996); although the data presented by Milan et al. (1997), as well as in this paper, suggest that cell death may provide an additional mechanism of size control.
In this paper, we demonstrate that ectopic expression of activated Ras1 (Ras1V12) during imaginal disc development is sufficient to drive cell proliferation and causes hyperplastic growth of imaginal discs. Further, Ras1V12 induces hyperplasia of both the eye imaginal disc and the adult eye when expressed under the control of an eyeless enhancer, which is an early eye-specific enhancer element. In addition, activated Ras1 induces non-autonomous cell death in the imaginal discs and, at high expression levels, can lead to the complete ablation of the adult eye. We also show that the MAPK cascade is required for RAS1 mitogenic response since loss-of-function mutations in genes encoding RAF, MEK, MAPK and KSR dominantly suppress hyperplastic growth, as do mutations in the RAS1 effector loop that disrupt the RAS-RAF interaction. Two of these RAS1 effector loop mutants can, however, act synergistically with a gain-of-function mutation in MAPK, which raises the possibility that alternate pathway(s) downstream of RAS1 may also be required.
MATERIALS AND METHODS
Fly stocks
Flies carrying UAS-Ras1wt, UAS-Ras1V12, UAS-Ras1V12 S35, UAS-Ras1V12 G37, UAS-Ras1V12 C40 and eyEh-Ras1V12 were generated by P-element transformation. Flies carrying the dpp-GAL4 transgene (lines 40C.6 and 57A) are described in Staehling-Hampton et al. (1994). The wg-lacZ and hh-lacZ enhancer trap lines are described in Kassis et al. (1992) and Lee et al. (1992), respectively. Sevenmaker mutant flies are described in Brunner et al. (1994). The strains wg-lacZ; dpp-GAL4(40C.6)/Tl(2;3) SM6-TM6B and dpp-GAL4(57A); hh-lacZ/Tl(2;3) SM6-TM6B were kindly provided by Jessica Treisman. The UAS-P21 lines and the UAS-P35 lines were generous gifts from Iswar Hariharan and Bruce Hay, respectively.
Construction of the UAS-Ras1 and P[eyEh-Ras1V12]
The Ras1wt and Ras1V12 genomic fragments were cut out of Bluescript vector (Fortini et al., 1992) as KpnI (blunt-ended)/NotI fragments and ligated into the pUAST vector at EcoRI (blunt-ended)/NotI. The Ras1 effector loop mutants were generated by PCR-mediated in vitro mutagenesis and confirmed by sequencing. Each effector loop mutant fragment (EcoRI/BglII) was combined in a three-way ligation with a BglII/NotI fragment encoding the remainder of the Ras1 sequences and EcoRI/NotI digested pUAST vector.
The 3.4 kb EcoRI genomic enhancer fragment from the eyeless gene (a generous gift from U. Walldorf and W. Gerhing) was blunt-end ligated into the XhoI site upstream of the hsp70 promoter of P-CaSpeR-heat shock (Thummel et al., 1988). The Ras1V12 genomic fragment was cut out of pBS vector (Fortini et al., 1992) as ClaI (blunt-ended)/NotI and ligated into the P[eyEh] vector at HpaI/NotI.
Histology
UAS-lacZ, hh-lacZ and wg-lacZ expression patterns in wing imaginal discs were detected by X-gal staining as described by Brand and Perrimon (1993). BrdU labeling was done following a modified protocol described by Truman and Bate (1988). Third instar wing discs were dissected in phosphate buffer and incubated in a 60 μg/ml solution of BrdU in phosphate buffer for 30 minutes at 25°C. Acridine orange staining of third instar larval wing discs was done as described in Spreij (1971). Flies were prepared for scanning electron microscopy as described in Kimmel et al. (1990).
RESULTS
Expression of Ras1V12 induces ectopic cell proliferation and hyperplastic growth of imaginal discs
To determine whether Ras1 activation is sufficient to drive cell proliferation, we used the GAL4-UAS system (Brand and Perrimon, 1993) to express in imaginal discs both the wild-type form of Ras1 (UAS-Ras1wt) and a mutationally activated form of Ras1, in which valine is substituted for glycine at position 12 (UAS-Ras1V12). In these studies, transgenic flies carrying either UAS-Ras1wt or UAS-Ras1V12 (see Materials and Methods) were crossed to a fly strain in which GAL4 was expressed under the control of the decapentaplegic (dpp) disc enhancer fragment (dpp-GAL4; Staehling-Hampton et al., 1994). dpp-GAL4 was chosen because it drives expression of GAL4 in all of the imaginal discs throughout development, and in only a limited portion of each disc resembling the pattern of endogenous dpp expression, as shown by UAS-lacZ (Fig. 1A).
Expression of activated Ras1 induces ectopic cell proliferation and hyperplastic growth in imaginal discs. Third instar wing discs were stained for β-galactosidase activity (blue; A-H,K,L), and with anti-BrdU antibody (brown; I,J). All discs are shown with anterior to the left. All tissues were isolated from animals raised at 25°C, except for the one shown in G, which was isolated from an animal raised at 18°C. In wing imaginal discs, dpp-GAL4 expression, and therefore expression of UAS-Ras1wt and UAS-Ras1V12, occurs in a stripe along the anterior-posterior compartment border (A/P border) just on the anterior side, as visualized by UAS-lacZ staining (A,B). hh-lacZ is expressed exclusively in the posterior compartment (C,D), whereas wg-lacZ is expressed along the wing margin and surrounding the wing pouch (E-H,K,L). The asterisks (H) show the approximate location of the A/P border. The genotypes are as follows: (A) UAS-Ras1wt/+; dpp-GAL4 /UAS-lacZ; (B) UAS-Ras1V12/+; dpp-GAL4 /UAS-lacZ; (C) UAS-Ras1wt/dpp-GAL4 /+; hh-lacZ/+; (D) UAS-Ras1V12/dpp-GAL4; hh-lacZ/+; (E,I)UAS-Ras1wt/wg-lacZ; dpp-GAL4 /+; (F,G,J)UAS-Ras1V12/wg-lacZ; dpp-GAL4 /+; (H) UAS-Ras1V12/wg-lacZ; dpp-GAL4 /UAS-Ras1V12; (K) wg-lacZ/+; dpp-GAL4/UAS-P21; (L) UAS-Ras1V12/wg-lacZ; dpp-GAL4 /UAS-P21.
Expression of activated Ras1 induces ectopic cell proliferation and hyperplastic growth in imaginal discs. Third instar wing discs were stained for β-galactosidase activity (blue; A-H,K,L), and with anti-BrdU antibody (brown; I,J). All discs are shown with anterior to the left. All tissues were isolated from animals raised at 25°C, except for the one shown in G, which was isolated from an animal raised at 18°C. In wing imaginal discs, dpp-GAL4 expression, and therefore expression of UAS-Ras1wt and UAS-Ras1V12, occurs in a stripe along the anterior-posterior compartment border (A/P border) just on the anterior side, as visualized by UAS-lacZ staining (A,B). hh-lacZ is expressed exclusively in the posterior compartment (C,D), whereas wg-lacZ is expressed along the wing margin and surrounding the wing pouch (E-H,K,L). The asterisks (H) show the approximate location of the A/P border. The genotypes are as follows: (A) UAS-Ras1wt/+; dpp-GAL4 /UAS-lacZ; (B) UAS-Ras1V12/+; dpp-GAL4 /UAS-lacZ; (C) UAS-Ras1wt/dpp-GAL4 /+; hh-lacZ/+; (D) UAS-Ras1V12/dpp-GAL4; hh-lacZ/+; (E,I)UAS-Ras1wt/wg-lacZ; dpp-GAL4 /+; (F,G,J)UAS-Ras1V12/wg-lacZ; dpp-GAL4 /+; (H) UAS-Ras1V12/wg-lacZ; dpp-GAL4 /UAS-Ras1V12; (K) wg-lacZ/+; dpp-GAL4/UAS-P21; (L) UAS-Ras1V12/wg-lacZ; dpp-GAL4 /UAS-P21.
The UAS-lacZ-staining pattern in UAS-Ras1wt/UAS-lacZ/dpp-GAL4 wing discs (Fig. 1A) is indistinguishable from discs carrying only UAS-lacZ and dpp-GAL4 (data not shown). In UAS-Ras1V12/UAS-lacZ/dpp-GAL4 wing discs lacZ staining is distorted and more diffuse (Fig. 1B), and the discs are often slightly larger than normal.
Two enhancer trap lines, hedgehog (hh-lacZ) and wingless (wg-lacZ), were used to examine the effects that ectopic expression of Ras1wt and Ras1V12 have on wing disc morphology and patterned gene expression. UAS-Ras1wt/dpp-GAL4 has no effect on either hh-lacZ or wg-lacZ-staining patterns (Fig. 1C and E). On the contrary, UAS-Ras1V12/dpp-Gal4 causes severe disruptions in the normal folding pattern on the anterior side of the A/P border, while the folding pattern in the posterior compartment appears to be relatively unaffected (Fig. 1D). Noteworthy are the appearance of new folds not normally present in wild-type discs, particularly along the A/P border. UAS-Ras1V12 disrupts wg-lacZ expression on both the dorsal and ventral edges of the wing pouch at the A/P border where dpp-GAL4 drives its expression (see arrowheads in Fig. 1F). Along the wing margin, wg-lacZ staining is unbroken, although it is somewhat distorted and more diffuse in the region of UAS-Ras1V12 expression. Disruption of wg-lacZ staining does not appear to be due to loss of expression but rather appears to result from displacement of the wg-expressing cells, such that they now abut the A/P border (see arrowheads in Fig. 1F). Alterations of both the folding pattern and wg-lacZ expression appear to be caused by ectopic tissue growth induced by Ras1V12.
The effects of UAS-Ras1V12 expression are temperature-sensitive and dose-sensitive. For example, UAS-Ras1V12/dpp-GAL4 animals raised at 18°C rather than 25°C show less severe disruptions of wg-lacZ staining (see arrowheads in Fig. 1G) and instead of dying as early pupae die as young adults with severe developmental defects (see Fig. 2B and D). Two copies of UAS-Ras1V12 at 25°C cause more severe disruption of wg-lacZ staining, including the staining along the wing margin (Fig. 1H). Also, the anterior compartment of the wing disc is dramatically overgrown, while the posterior compartment is more compacted (Fig. 1H, see asterisk). Similar Ras1V12-dependent overgrowth was generated in the posterior compartment when an engrailed-GAL4 line was used (data not shown). Under each of these conditions, the discs appear to remain as monolayer epithelia (unpublished observations).
Activated Ras1 induces non-autonomous cell death in imaginal discs. Adult legs (A,B) and wings (C-H) are shown with anterior up. The longitudinal wing veins (L1-L5), as well as the anterior (acv) and posterior cross veins (pcv) are labeled (C). Acridine orange staining of third instar wing discs reveals dying cells (I-L). Note, the brightly stained region in I is a piece of trachea. UAS-Ras1V12/dpp-GAL4 animals raised at 18°C die as young adults with mild rough eyes (data not shown), severely malformed legs with missing or fused tarsal segments (see brackets in B) and smaller wings with broad L3 wing vein (see arrowhead in D). The genotypes are as follows: (A,C,I) UAS-Ras1wt/+; dpp-GAL4 /+; (B,D,J) UAS-Ras1V12/+; dpp-GAL4/+; (E,G) UAS-P21/+; +/+; dpp-GAL4/UAS-P21; (F,H) UAS-P21/+; UAS-Ras1V12/+; dpp-GAL4/UAS-P21; (K) UAS-Ras1V12/+; dpp-GAL4/UAS-Ras1V12; (L) UAS-Ras1V12/+; dpp-GAL4/UAS-P35.
Activated Ras1 induces non-autonomous cell death in imaginal discs. Adult legs (A,B) and wings (C-H) are shown with anterior up. The longitudinal wing veins (L1-L5), as well as the anterior (acv) and posterior cross veins (pcv) are labeled (C). Acridine orange staining of third instar wing discs reveals dying cells (I-L). Note, the brightly stained region in I is a piece of trachea. UAS-Ras1V12/dpp-GAL4 animals raised at 18°C die as young adults with mild rough eyes (data not shown), severely malformed legs with missing or fused tarsal segments (see brackets in B) and smaller wings with broad L3 wing vein (see arrowhead in D). The genotypes are as follows: (A,C,I) UAS-Ras1wt/+; dpp-GAL4 /+; (B,D,J) UAS-Ras1V12/+; dpp-GAL4/+; (E,G) UAS-P21/+; +/+; dpp-GAL4/UAS-P21; (F,H) UAS-P21/+; UAS-Ras1V12/+; dpp-GAL4/UAS-P21; (K) UAS-Ras1V12/+; dpp-GAL4/UAS-Ras1V12; (L) UAS-Ras1V12/+; dpp-GAL4/UAS-P35.
To determine whether Ras1V12 induces cell proliferation, third instar wing discs were labeled with BrdU, which is incorporated by cells during S-phase. UAS-Ras1wt/dpp-GAL4 wing discs have a uniform pattern of cell division, which is indistinguishable from wild-type (Fig. 1I; see also Schubiger and Palka, 1987; Milan et al., 1996). In contrast, the UAS-Ras1V12/dpp-GAL4 wing discs have a dramatic increase in BrdU incorporation in a broad band along the A/P border, corresponding to the domain of dpp-GAL4 expression (see brackets in Fig. 1J). No increase in BrdU incorporation is seen in the posterior compartment or the regions just anterior to this band of high BrdU incorporation. Thus, ectopic expression of Ras1V12 under control of dpp-GAL4 induces cell proliferation, which leads to hyperplastic growth that disrupts both the normal folding and patterned gene expression in the anterior compartment.
Human cyclin-dependent kinase inhibitor, P21CIP1/WAF1 suppresses Ras1V12-dependent cell proliferation
We attempted to suppress Ras1V12-induced cell proliferation by co-expressing the human cyclin-dependent kinase inhibitor, P21CIP1/WAF1 (UAS-P21; Harper et al., 1993; El-Deiry et al., 1993). Ectopic expression of human P21 has been successfully used to block the second mitotic wave in the developing eye imaginal disc (de Nooij and Hariharan, 1995). UAS-P21/dpp-GAL4 wing discs are morphologically normal in size and shape (Fig. 1K). At 25°C, co-expression of UAS-P21 appears to suppress most, but not all, of the UAS-Ras1V12-induced cell proliferation and hyperplastic growth seen in wing discs (see arrowhead in Fig. 1L). Under these conditions, however, one copy of UAS-P21 is not sufficient to suppress the lethality but does delay the lethal phase, similar to what is seen in UAS-Ras1V12/dpp-GAL4 flies raised at 18°C (data not shown).
Flies carrying two copies of UAS-P21 and dpp-GAL4 raised at 18°C are morphologically normal, except for loss of the anterior cross vein and part of the L3 wing vein (see arrowhead in Fig. 2E,G). In addition, cells in the L3-L4 intervein region are larger than normal as visualized by the more widely spaced wing hairs (see arrowhead in Fig. 2E,G). At 18°C, co-expression of two copies of UAS-P21 with UAS-Ras1V12 results in the complete suppression of lethality associated with activated Ras1 expression. Morphologically, the wing is restored to its normal shape and size, although UAS-P21 only partially suppresses the extra wing vein material along L3 (see arrowhead in Fig. 2F,H). Interestingly, expression of UAS-Ras1V12 completely suppresses the larger cell size caused by UAS-P21 expression (Fig. 2H).
Ras1V12 induces non-autonomous cell death
As noted earlier, UAS-Ras1V12/dpp-GAL4 animals raised at 18°C die as young adults with deletions of certain structures, particularly those arising from cells adjacent to the domain where activated Ras1 is expressed, e.g., the tarsal segments of the leg (Fig. 2B,D). To address whether these deletions result from UAS-Ras1V12-induced cell death, we stained third et al., 1994). Cell death is greatly reduced in the region where dpp-GAL4 drives co-expression of UAS-Ras1V12 and UAS-P35 but is still considerable in other regions of the disc, especially in its anterior portion (see brackets in Fig. 2L). When raised at 18°C, UAS-Ras1V12/UAS-P35/dpp-GAL4 animals still die as young adults with deletions of structures (data not shown), instar wing imaginal discs with acridine orange. Very little, if any, cell death is observed in either wild-type (data not shown) or UAS-Ras1wt/dpp-GAL4 wing discs (Fig. 2I). Wing discs isolated from animals carrying one or two copies of UAS-Ras1V12, however, have a marked increase in cell death (Fig. 2J,K) that includes but is not limited to the region where Ras1V12 is expressed. This suggests that ectopic expression of Ras1V12 induces cell death non-autonomously. To confirm this non-autonomy, the baculovirus cell death inhibitor P35 and Ras1V12 were co-expressed (UAS-P35; B. Hay, personal communication; also see Hay presumably as a consequence of this non-autonomous cell death.
The MAP kinase cascade is required for Ras1V12-dependent cell proliferation
The RAF serine/threonine kinase, which is the best characterized Ras effector, is recruited to the plasma membrane upon Ras activation via a direct protein-protein interaction between the effector loop of RAS and the Ras-binding domain within the N-terminal domain of RAF (reviewed in Morrison and Cutler, 1997; Katz and McCormick, 1997). A dominant-negative Raf allele, RafXS-2541, was tested for its ability to suppress as a heterozygote (i.e., dominantly) the Ras1V12-induced proliferation. At 25°C, RafXS-2541/+ dominantly suppresses most of the hyperplastic growth seen in UAS-Ras1V12/dpp-GAL4 wing discs (see arrowheads in Fig. 3B). Similarly, loss-of-function mutations in the genes encoding the Drosophila homologues of MEK, MAPK and KSR dominantly suppress most of the Ras1V12-induced proliferation (data not shown), consistent with the MAPK cascade playing a critical downstream role during cell proliferation. Conversely, a dominant gain-of-function mutation in Drosophila MAPK, Sevenmaker (Sem, rolledSem, MAPKSem; Brunner et al., 1994), dominantly enhances activated Ras1 overproliferation, resulting in a much larger disc with severe disruptions of morphology (Fig. 3C).
The Raf/MAP kinase cascade is necessary but may not be sufficient for RASV12-induced proliferation. Third instar wing discs (anterior is to the left) were stained for β-galactosidase activity to reveal the wg-lacZ expression pattern (blue; A,B,D-G; the wing disc shown in C does not carry the wg-lacZ enhancer trap line). All wing discs are displayed at the same magnification. The genotypes are as follows: (A) UAS-Ras1V12/wg-lacZ; dpp-GAL4 /+; (B) RafXS-2541/+; UAS-Ras1V12/wg-lacZ; dpp-GAL4 /+; (C) UAS-Ras1V12/MAPKSem; dpp-GAL4 /+; (D) UAS-Ras1V12 S35/wg-lacZ; dpp-GAL4 /+; (E) UAS-Ras1V12 G37/+; wg-lacZ/+; dpp-GAL4/+; (F) UAS-Ras1V12 C40/+; wg-lacZ/+; dpp-GAL4 /+; (G) UAS-Ras1V12 G37/+; MAPKSem/wg-lacZ; dpp-GAL4/+.
The Raf/MAP kinase cascade is necessary but may not be sufficient for RASV12-induced proliferation. Third instar wing discs (anterior is to the left) were stained for β-galactosidase activity to reveal the wg-lacZ expression pattern (blue; A,B,D-G; the wing disc shown in C does not carry the wg-lacZ enhancer trap line). All wing discs are displayed at the same magnification. The genotypes are as follows: (A) UAS-Ras1V12/wg-lacZ; dpp-GAL4 /+; (B) RafXS-2541/+; UAS-Ras1V12/wg-lacZ; dpp-GAL4 /+; (C) UAS-Ras1V12/MAPKSem; dpp-GAL4 /+; (D) UAS-Ras1V12 S35/wg-lacZ; dpp-GAL4 /+; (E) UAS-Ras1V12 G37/+; wg-lacZ/+; dpp-GAL4/+; (F) UAS-Ras1V12 C40/+; wg-lacZ/+; dpp-GAL4 /+; (G) UAS-Ras1V12 G37/+; MAPKSem/wg-lacZ; dpp-GAL4/+.
Although RafXS-2541/+ suppresses most of the overproliferation seen at 25°C, it only partly suppresses the cell death and lethality associated with UAS-Ras1V12/dpp-GAL4. The surviving adults have severe developmental defects, similar to UAS-Ras1V12/dpp-GAL4 animals raised at 18°C (data not shown). The residual hyperplastic growth seen in RafXS-2541/UAS-Ras1V12/dpp-GAL4 animals may reflect residual RAF activity present in RafXS-2541/+ heterozygotes. Since animals that are hemizygous or homozygous mutant for a null allele of Raf (Raf11-29) die as third instar larvae lacking imaginal discs (Nishida et al., 1988), we tested to see whether expression of activated Ras1 can bypass a requirement for Raf. Indeed, no imaginal disc proliferation is seen in hemizygous Raf11-29 males in which UAS-Ras1V12 is expressed under control of dpp-GAL4 (data not shown), consistent with a requirement for Raf and the MAPK cascade.
RAS1 may also act through signaling pathway(s) distinct from the MAPK cascade
Although Raf is required for Ras1-dependent proliferation, it may not be sufficient for mediating all Ras1 signaling. In mammalian cells, RAS not only interacts with RAF but can also interact with effectors distinct from RAF, some of which act independently or in parallel to RAF (reviewed in Katz and McCormick, 1997). Three different Ras effector loop mutants (T35S, E37G, Y40C) that block or impair the ability of RAS to bind RAF have been used to distinguish between RAF and alternate effector pathways downstream of mammalian RAS (White et al., 1995; Khosravai-Far et al., 1996; Joneson et al., 1996). These effector loop mutants can act synergistically with each other to transform mammalian cells, which suggests that they differentially affect parallel downstream pathways (White et al., 1995; Joneson et al., 1996; Khosravai-Far et al., 1996).
Based on these observations and the conservation of RAS1 effector loop sequence, we generated three different effector loop mutants (T35S, E37G and Y40C) in the activated form of Drosophila Ras1 (Ras1V12 S35, Ras1V12 G37 and Ras1V12 C40). Unlike mammalian cells, in which H-RASV12 S35 is attenuated in its ability to transform cells, UAS-Ras1V12 S35/dppGAL4 produces approximately the same amount of hyperplastic growth in third instar wing disc as UAS-Ras1V12 (see arrowhead in Fig. 3D). In contrast, both UAS-Ras1V12 G37 and UAS-Ras1V12 C40 mutants display little, if any, hyperplastic growth in wing discs (Fig. 3E,F, respectively) In both cases, animals survive to adulthood but have minor differentiation defects, such as mild rough eyes (data not shown) and extra wing vein material in the case of UAS-Ras1V12 G37/dpp-GAL4 but not in UAS-Ras1V12 C40/dpp-GAL4 (see arrowheads in Fig. 4B,C). UAS-Ras1V12 G37 and UAS-Ras1V12 C40 do not act synergistically and produce no ectopic proliferation in the imaginal discs and the differentiation defects seen in the adults appear to be additive rather than synergistic (data not shown). These results confirm that activation of the MAPK cascade is an essential downstream event during RAS1-induced proliferation.
The E37G and Y40C Ras effector loop mutants block RAS1V12-induced proliferation but retain some residual activity indicative of an additional downstream pathway parallel to RAF. Adult wings (A-F) are shown with anterior up. The genotypes are as follows: (A) wild type; (B) UAS-Ras1V12 G37/+; dpp-GAL4/+; (C) UAS-Ras1V12 C40/+; dpp-GAL4 /+; (D) MAPKSem/+; (E) UAS-Ras1V12 C40/+; MAPKSem/+; dpp-GAL4/+; (F) UAS-Ras1V12 C40/RafXS-2541; MAPKSem/+; dpp-GAL4/+.
The E37G and Y40C Ras effector loop mutants block RAS1V12-induced proliferation but retain some residual activity indicative of an additional downstream pathway parallel to RAF. Adult wings (A-F) are shown with anterior up. The genotypes are as follows: (A) wild type; (B) UAS-Ras1V12 G37/+; dpp-GAL4/+; (C) UAS-Ras1V12 C40/+; dpp-GAL4 /+; (D) MAPKSem/+; (E) UAS-Ras1V12 C40/+; MAPKSem/+; dpp-GAL4/+; (F) UAS-Ras1V12 C40/RafXS-2541; MAPKSem/+; dpp-GAL4/+.
The mild differentiation defects associated with both RAS1V12 G37 and RAS1V12 C40 indicate that these mutant proteins retain some signaling activity, which might result from either a residual capacity to bind and activate RAF, or from signal transmission via a Raf-independent pathway. Under either scenario, one might expect these effector loop mutants to act synergistically with MAPKSem, the MAPK gain-of-function mutation. MAPKSem flies display no apparent proliferation defect on their own. The adults have moderately rough eyes and extra wing vein material (Brunner et al., 1994), although extra vein material is only occasionally seen along the L3 wing vein (see arrowhead in Fig. 4D). In fact, UAS-Ras1V12 G37/dpp-GAL4 and MAPKSem do act synergistically to produce a small amount of hyperplastic growth in third instar wing discs (see arrowhead in Fig. 3G), but only a fraction of that seen with UAS-Ras1V12/dpp-GAL4 at 25°C. These animals die as pharate adults at 25°C and have deletions of tarsal segments and extra wing vein material along L3, similar to UAS-Ras1V12/dpp-GAL4 animals raised at 18°C (data not shown). No apparent ectopic proliferation is seen in imaginal discs of animals carrying MAPKSem and UAS-Ras1V12 C40/dpp-GAL4 (data not shown). The adults, however, often have wing blisters and extra vein material along the L3 vein not normally seen in either MAPKSem or UAS-Ras1V12 C40/dpp-GAL4 flies alone (see arrowhead in Fig. 4E). If the synergistic activity between these effector loop mutants and MAPKSem is a consequence of residual RAS-RAF interaction, then it should be dramatically suppressed by the dominant-negative Raf allele. In fact, the extra wing vein material and blisters seen in MAPKSem/UAS-Ras1V12 C40/dpp-GAL4 flies are not dramatically suppressed by RafXS-2541/+ (see arrowhead in Fig. 4F), consistent with RAS1 acting via a pathway distinct from the MAPK cascade.
Expression of Ras1V12 under control of an eyeless-enhancer construct causes hyperplastic growth of both the eye imaginal disc and the adult eye
To facilitate future genetic screens, we wished to generate a stable transgenic fly strain in which Ras1V12-induced hyperproliferation does not lead to lethality. Expression of activated Ras1 exclusively in the eye imaginal disc throughout development would be ideal, since the eye is dispensable for viability and fertility and, unlike the wing, is easily scored in adults that fail to eclose. To achieve this, we used a 3.4 kb genomic enhancer fragment from the eyeless (ey) gene that has been shown to confer expression of a lacZ reporter throughout eye development (Walsdorf and Gehring, personal communications). Eye imaginal discs from eyEh-Ras1V12 third instar larvae raised at 18°C are larger than normal (Fig. 5G), consistent with ectopic cell proliferation and hyperplastic growth, but the adult eyes are rough and slightly smaller than normal (Fig. 5B). When raised at 25°C, the eye imaginal discs from eyEh-Ras1V12 animals are dramatically larger than wild-type (Fig. 5H). These animals survive as adults and often have larger-than-normal eyes, with protruding outgrowths of eye tissue and cuticle emanating from the posterior margin of the eye (Fig. 5C). Occasionally, rather than having larger eyes, ventral portions or all of one eye may be missing (data not shown). When raised at 29°C, eyEh-Ras1V12 larvae have no eye-antennal discs (data not shown). These animals die as pharate adults and completely lack all head structures derived from the eye-antennal disc. Only the proboscis, which develops from the labial disc, is seen protruding from the thorax where the head is normally located (see arrowhead in Fig. 5D).
Expression of RasV12 under the control of an eyeless enhancer element (eyEh-Ras1V12) triggers hyperplastic growth in the eye imaginal disc and the adult eye. Scanning electron micrographs of adult eyes (A-E) and eye-antennal imaginal discs (F-H) are shown with anterior to the left. The animal in panel D has no head, only the proboscisis present (arrowhead). The genotypes are as follows: (A,F) wild type; (B,G) eyEh-Ras1V12 raised at 18°C; (C,H) eyEh-Ras1V12 raised at 25°C; (D) eyEh-Ras1V12 raised at 29°C; (E) RafXS-2541/+; eyEh-Ras1V12 raised at 29°C.
Expression of RasV12 under the control of an eyeless enhancer element (eyEh-Ras1V12) triggers hyperplastic growth in the eye imaginal disc and the adult eye. Scanning electron micrographs of adult eyes (A-E) and eye-antennal imaginal discs (F-H) are shown with anterior to the left. The animal in panel D has no head, only the proboscisis present (arrowhead). The genotypes are as follows: (A,F) wild type; (B,G) eyEh-Ras1V12 raised at 18°C; (C,H) eyEh-Ras1V12 raised at 25°C; (D) eyEh-Ras1V12 raised at 29°C; (E) RafXS-2541/+; eyEh-Ras1V12 raised at 29°C.
We crossed mutants that suppressed or enhanced Ras1V12-induced differentiation effects (Karim et al., 1996) to eyEh-Ras1V12 at 18°C, 25°C and 29°C. Loss-of-function mutations in Raf, MEK, MAPK and Ksr dominantly suppress both the lethality and disc overgrowth seen at all three temperatures, although the mutations suppress these phenotypes to varying degrees and none completely at 29°C. For example, RafXS-2541 dominantly suppresses eyEh-Ras1V12 at 29°C such that most animals have head structures and some survive to adulthood with a limited amount of hyperplasia (Fig. 5E). In addition, several enhancers isolated in the sev-Ras1V12 screen also enhance eyEh-Ras1V12 such that the animals die at 18°C. These are usually missing most or all head structures, similar to eyEh-Ras1V12 at 29°C (data not shown).
DISCUSSION
In this paper, we demonstrate that expression of mutationally activated Ras1 (Ras1V12) induces ectopic cell proliferation, hyperplastic growth and non-autonomous cell death during imaginal disc development.
When dpp-GAL4 is used to drive UAS-Ras1V12 expression in the wing disc, ectopic cell proliferation occurs in a stripe along the anterior side of the A/P border. Dramatic expansion of the disc profile occurs with high levels of activated Ras1 activity (e.g., with two copies of UAS-Ras1V12, or in combination with the MAPKSem), yet the growth appears to be hyperplastic rather than neoplastic since the disc remains a monolayer epithelium and is not invasive (data not shown).
Two types of overgrowth phenotypes are observed in Drosophila imaginal discs, hyperplastic and neoplastic. Neoplastic growth of imaginal discs can be characterized by the loss of its monolayer epithelial character and its inability to differentiate. In addition, neoplastic growths are invasive when transplanted into normal hosts. Hyperplastic growths retain their monolayer epithelial structure and their ability to differentiate, but are not invasive (reviewed in Gateff, 1994; Bryant et al., 1993). Hyperplastic growths can be further subdivided into two classes; those that result as a consequence of a pattern duplication (e.g., the overgrowth that results from ectopic dpp expression; reviewed in Edgar and Lehner, 1996), and those that display no pattern duplication but instead result from a general loss of growth control (e.g., as seen in warts or fat loss-of-function mutants; reviewed in Bryant et al., 1993). RAS1V12-induced hyperplastic growth displays no obvious pattern duplication in the discs or adult structures, but instead displays larger than normal tissues (Fig. 5C,H).
These results demonstrate that Ras1 plays an important role in the control of imaginal disc cell proliferation; however, it is unclear how Ras1 signaling is integrated with signaling pathways that control pattern formation.
Is non-autonomous cell death a manifestation of cell competition?
Ras1V12 animals that survive to adulthood often have deletions of adult structures (e.g., deletions of tarsal segments in UAS-Ras1V12/dpp-GAL4 animals and head structures in eyEh-Ras1V12 animals). In each case, the structures most commonly deleted include those that arise from cells that lie adjacent to the Ras1V12-expressing cells within the imaginal disc.
Presumably the loss of wild-type cells results from induction of non-autonomous cell death. It is, however, unclear whether UAS-Ras1V12-expressing cells trigger the non-autonomous cell death by sending a lateral signal, or whether this occurs indirectly as a consequence of excess proliferation in the disc. Consistent with this latter possibility, co-expression of two copies of P21, which should block cell proliferation but not affect a lateral signal, suppresses the deletion of structures in the adult (Fig. 2F, data not shown). Complementary to our results, which indicate that ectopic proliferation leads to the induction of non-autonomous cell death, Milan and Garcia-Bellido (1997) have shown that experimentally induced cell death in one region of the disc triggers non-autonomous cell death, as well as cell proliferation.
One interpretation is that non-autonomous cell death is the manifestation of cell competition. The notion of cell competition was first proposed when it was observed that clones of Minute/+ mutant cells, which have a cell-autonomous slower rate of division, could not be recovered when generated in non-Minute (+/+) animals, even though heterozygous mutants are not otherwise cell lethal (Morata and Ripoll, 1977). Additionally, when homozygous wild-type clonal tissue was generated in Minute/+ mutant animals, they grew to be very large at the expense of Minute mutant cells, hence, the term cell competition. In the Minute experiments, cell competition is restricted by compartment boundaries, and mosaic compartments are always of normal size and shape. This is not true in the Ras1V12 experiments since the anterior compartment of the wing disc is usually bigger in UAS-Ras1V12/dpp-GAL4 animals. In this case, Ras1V12-expressing cells might ‘compete’ with neighboring non-expressing cells and induce non-autonomous cell death. This cell death, however, is insufficient to completely compensate for the increased proliferation, so the discs are not normal size or shape. Increased cell death may represent a mechanism to compensate for excessive proliferation and regulate the overall disc size. Consequently, the final outcome of Ras1V12 expression on adult structures appears to be a balance between the amount of proliferation versus cell death.
The MAPK cascade is necessary for Ras-induced proliferation
Our results raise important questions about the nature of Ras1 signaling during cell proliferation and how it differs during cell differentiation. Understanding Ras signaling during oncogenesis requires a better understanding of its downstream effector(s). Two lines of evidence demonstrate that the MAPK cascade is necessary for Ras1-dependent proliferation: First, loss-of-function mutations in Raf, MEK, MAPK and Ksr genes strongly suppress of RAS1V12-induced hyperplasia. Second, Ras effector loop mutants (E37G and Y40C) that disrupt the RAS-RAF interaction completely suppress RAS1V12-induced proliferation.
Even though the RAS1 effector loop and the Ras-binding domain of Drosophila RAF are conserved between flies and mammals, the RAS1 effector loop mutants reveal some differences between RAS1 and mammalian RAS. For example, the T35S effector loop mutant, which impairs but does not block H-RAS-RAF interaction in the yeast two-hybrid system and blocks mammalian cell transformation, does not significantly alter the ability of activated RAS1 to drive cell proliferation. H-RASV12 G37 and H-RASV12 C40 are impaired in their abilities to stimulate the MAPK cascade (White et al., 1995; Khosravai-Far et al., 1996), but both retain a limited ability to transform certain mammalian cell lines, suggestive of Raf-independent pathways, and together they can act synergistically to more effectively cause transformation. UAS-Ras1V12 G37 and UAS-Ras1V12 C40 do not produce ectopic cell proliferation on their own, but do produce mild differentiation defects indicative of some residual activity. Furthermore, RAS1V12 G37 and RAS1V12 C40 do not have complementary activities, since co-expression produces additive rather than synergistic phenotypes, although both can act synergistically with MAPKSem, a gain-of-function allele. This synergistic activity might result from MAPKSem augmenting a low level activity generated by a residual capacity of RAS1V12 G37 or RAS1V12 C40 to bind and activate RAF. Alternatively, RAS1V12 G37 and RAS1V12 C40 might retain signaling activity via a pathway parallel to RAF, and this is augmented by MAPKSem gain-of-function activity. The former does not seem likely since RafXS-2541/+ does not dominantly suppress the extra wing vein material in either UAS-Ras1V12 G37/dpp-GAL4 (data not shown) or Sem/UAS-Ras1V12 C40/dppGAL4 flies (Fig. 4F). These results are instead consistent with an alternate signaling pathway(s) functioning downstream of RAS1 but parallel to the MAPK cascade.
Ras signal transduction during cell proliferation
Mutationally activated H-RAS is implicated in the development of some human cancers, which is born out by its ability to transform certain mammalian cell lines. Paradoxically, H-RASV12 triggers senescence in human and rodent primary cell lines (Serrano et al., 1997). These different responses presumably reflect the differing mutational histories of these cell lines, which can lead to confusing or contradictory conclusions regarding Ras function. The high degree of conservation of Ras signaling pathways between Drosophila and humans demonstrates that flies provide a good in vivo system for understanding Ras-dependent proliferation. Furthermore, both fly and mammalian cells require Ras for proliferation, and activated Ras is sufficient to induce proliferation. This does not appear to be true of all metazoan organisms since, in C. elegans, Ras does not appear to play a role in proliferation (Yochem et al., 1997). RAS1V12-induced proliferation generated with UAS-Ras1V12 and eyEh-Ras1V12 combined with the genetic and molecular tools available in flies provide a powerful in vivo model system for addressing the role of Ras during cell proliferation, both in general and in a developmental context. Key points that this system can be used to address include the following. How does Ras activation lead to differentiation versus a mitogenic response? What signaling pathways in addition to the MAPK cascade contribute to transmission of the proliferation signal? What are the critical links between Ras signaling and the cell cycle machinery? What constrains the Ras1V12-dependent proliferation so that these cells maintain the monolayer?
The ability to address these questions within a developmental context should reveal how Ras signaling is integrated with the cell cycle machinery and with other signaling pathways during development and oncogenesis.
ACKNOWLEDGEMENT
We thank Jessica Treisman, Iswar Hariharan and Bruce Hay for fly stocks. We are also grateful to Michael Brodsky, Mary Craig and Kathleen Sullivan for critical comments on the manuscript, and the other members of the Rubin laboratory for helpful discussions and technical assistance. F. D. K. is funded with a Senior Postdoctoral Fellowship from the California Division of The American Cancer Society, with generous support from the Alameda County Cancer League.