Rescue of a Drosophila NF1 Mutant Phenotype by Protein Kinase A

See allHide authors and affiliations

Science  02 May 1997:
Vol. 276, Issue 5313, pp. 791-794
DOI: 10.1126/science.276.5313.791


The neurofibromatosis type 1 (NF1) tumor suppressor protein is thought to restrict cell proliferation by functioning as a Ras-specific guanosine triphosphatase–activating protein. However,Drosophila homozygous for null mutations of anNF1 homolog showed no obvious signs of perturbed Ras1-mediated signaling. Loss of NF1 resulted in a reduction in size of larvae, pupae, and adults. This size defect was not modified by manipulating Ras1 signaling but was restored by expression of activated adenosine 3′,5′-monophosphate–dependent protein kinase (PKA). Thus, NF1 and PKA appear to interact in a pathway that controls the overall growth of Drosophila.

The gene responsible for human NF1 encodes a large protein that contains a central domain related to Ras-specific guanosine triphosphatase–activating proteins (Ras-GAPs) (1). Although loss of NF1 expression correlates with increased Ras activity in several mammalian tumor cell types (2), it is not known which pathways are altered to produce the diverse symptoms observed in NF1 patients, which in addition to frequent benign and infrequent malignant tumors also include short stature and learning disabilities (1).

We identified a conserved Drosophila NF1 homolog (3). Comparison of 13,295 base pairs (bp) of genomic and 9750 bp of cDNA sequence showed that Drosophila NF1 consists of 17 constitutive and 2 alternatively spliced COOH-terminal exons. The two cDNAs predict proteins of 2764 and 2802 amino acids that are 60% identical to the human NF1 protein, neurofibromin (Fig.1). Sequence similarity is observed over the entire length of the proteins, including regions outside of the catalytic GAP-related domain (GRD) or the more extensive segment related to yeast inhibitor of RAS activity (IRA) proteins (4). No related sequences were identified during screens of several cDNA and genomic libraries, indicating that the identified gene may be the onlyDrosophila NF1 homolog. RNA in situ hybridization and staining of embryos and imaginal discs with monoclonal antibodies to the Drosophila protein (5) indicated thatNF1 is widely expressed in low amounts during all developmental stages (6). The Drosophila NF1 gene was mapped to cytogenetic interval 96F and subsequently localized to a 30-kb DNA segment between the bride of sevenless gene and the Enhancer of split [E(spl)] complex (7).

Figure 1

Drosophila NF1 gene structure and comparison of the encoded protein to human neurofibromin. (A) Intron-exon structure and location of translational start (AUG) and in-frame stop (*) codons. The arrow indicates the location of a P-element in NF1P2 . The extent of the deletion in NF1P1 is indicated by the line below the diagram. (B) Percentage amino acid sequence identity between the indicated segments ofDrosophila and human NF1. The GRD- and IRA-related segments are drawn as black and shaded boxes, respectively. (C) Alignment of Drosophila (top) and human NF1 proteins. Dashes were introduced to optimize the alignment. Downward pointing arrows indicate amino acids encoded by the last complete codon in each exon. The boxed segment shows the approximate extent of the GRD. Three positions where alternate splicing inserts short segments in human neurofibromin (28) are identified by upward pointing arrows. One of these locations corresponds exactly to the position where Drosophila exon 17 is joined to either exon 18a or 18b. Exon 18b includes a translational terminator after a single codon, and cDNAs harboring this exon predict a protein ending in PTDKAA. Eleven out of 17 Drosophila splice sites map within two codons of splice sites in the human NF1 gene (29). Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

To isolate mutant alleles at the NF1 locus, we mobilized a P -element transposon located within the E(spl)complex, about 15 kb downstream of NF1 (7). Among 1600 lines screened by inverse polymerase chain reaction (8), two showed evidence of de novo transposon insertions within the NF1 gene. One mutant allele,NF1P1 , has a deletion that removed all of theNF1 gene except for the first exon (Fig. 1). The deletion extends from the first NF1 intron to the site of the original P -element insertion and removes DNA encoding at least two E(spl) transcripts. The other allele,NF1P2 , contains a P -element in the first NF1 intron (Fig. 1). Neither allele expresses NF1 protein (Fig. 2A).

Figure 2

Size defect of NF1mutants: Rescue by expression of an hsNF1 transgene and evidence of reduced cell size. (A) Immunoblots of embryo extracts probed with monoclonal antibody to NF1 (antibody DNF-21). The arrow indicates a 280-kD immunoreactive protein that is present in the K33 parental strain but absent in both NF1P1 and NF1P2 . A transgenic strain harboring a second chromosome hsNF1 transgene had an increased amount of 280-kD immunoreactive protein even before heat shock (HS) induction. (B) Third instar larvae (top) and pupae (bottom) of the indicated genotypes. (C and D) Wings from K33 and NF1P1 flies, respectively. (E) Quantitation of the number of wing epidermal cells per square millimeter in the indicated strains.

Unlike Nf1-deficient mice (9), Drosophila NF1 mutants are viable and fertile. Although heterozygotes (NF1/+) had no obvious defects, homozygotes (NF1/NF1) of either allele were 20 to 25% smaller than flies of the parental K33 strain during all postembryonic stages. This growth defect was apparent under various culture conditions, and mutant animals did not display delayed eclosion or bristle phenotypes that are observed with several Minute mutations (10). The growth defect was fully rescued by expression of a heat shock–inducible hsNF1 (11) transgene (Fig. 2, A and B, and Table 1).

Table 1

Rescue of the pupal size defect of NF1mutants by hsNF1 and activated PKA, but not by activated Raf or by reduced Ras1 gene dosage. The size of pupae is given as the average plus or minus the standard deviation. With the Mann-Whitney rank-sum test (30), the following populations had statistically significant differences in median pupal lengths (P < 0.0001): K33 and NF1P1 , K33 and NF1P2 , NF1P1 andhsp70-PKA*; NF1P1 andNF1P2 and hsp70-PKA*;NF1P2 . The size differences betweenRas1e1B ; +/+ andRas1e1B ; NF1P2 , and between hsp70-PKA*; +/+ and hsp70PKA*;NF1P1 or hsp70-PKA*;NF1P2 pupae were significant at the same confidence level. Rel. size, relative size.

View this table:

To determine whether reduced cell proliferation or impaired cell growth underlies the smaller size of NF1 mutants, we compared the wings of wild-type and mutant animals. The linear dimensions of mutant wings were 20 to 25% smaller than those of wild-type flies (Fig. 2, C and D). Because each wing epidermal cell secretes a single hair, cell densities can be determined by counting the number of hairs in a defined region (12). Both homozygous mutants had a 30 to 35% higher cell density than flies of the parental line (Fig. 2E). To determine whether the reduced size of wing epidermal cells reflects a cell-autonomous defect, we used x-irradiation to induce mitotic recombination in the wing cells of heterozygous NF1 mutants (13). No difference in cell density was observed between multiple NF1−/− clones and surrounding tissue (6). The reduced size of the wing cells therefore reflects a nonautonomous requirement for NF1, perhaps reflecting a hormonal deficiency or impaired nutrition or metabolism. The eyes ofNF1 mutants showed a reduced number of ommatidia of normal size and structure (Fig. 3). Furthermore,NF1-deficient embryos were of normal size (Fig. 3). Thus, loss of NF1 affects the growth of various tissues in different ways.

Figure 3

Function of Drosophila NF1 as a Ras-GAP in vitro, but not as a major regulator of Torso- or Sevenless-mediated signaling in vivo. (A) Radioactivity remaining on H-Ras–[γ32P]guanosine triphosphate (GTP) or H-RasVal12–[γ32P]GTP after 10-min incubations with the indicated fusion proteins (h-NF1 is the human NF1 GRD; d-NF1 is the Drosophila GRD; WT, wild type). (B and C) RNA in situ hybridization, showing tailless expression. The embryos in (B) and (C) are K33 and NF1P2 homozygotes, respectively. (D) Retinal section of adult NF1P2 homozygotes, showing ommatidia of normal size and structure. Mutant eyes have only about 550 ommatidia, compared with ∼750 in K33 flies. The scale bar represents 10 μm.

The only known biochemical property of neurofibromin is that it negatively regulates Ras (2), but Drosophila NF1mutants did not exhibit phenotypic abnormalities associated with excess Ras1 or Ras2 activity. Whereas expression of activated Ras1or Ras2 transgenes results in widespread developmental defects (14), NF1 mutants are smaller but otherwise patterned normally. We therefore examined the regulation of Ras by NF1 both in vitro and in vivo. We assayed GAP activity of bacterial fusion proteins containing the catalytic domains of human p120GAP, or human or Drosophila NF1 (15). All three fusion proteins stimulated the guanosine triphosphatase (GTPase) activity of human H-Ras, but not of the constitutively active H-RasVal12 mutant (Fig. 3A).

We also examined Ras1 function in vivo. The DrosophilaRas1 protein functions in signaling pathways downstream of several receptor tyrosine kinases (RTKs), including Torso and Sevenless. Because minor perturbations in Ras1 function have phenotypic consequences in each of these pathways, we examined whether loss ofNF1 perturbed Torso-controlled specification of embryonic terminal structures or Sevenless-mediated photoreceptor differentiation. The pattern of tailless expression, which is regulated by Torso (16), was normal inNF1-deficient embryos (Fig. 3, B and C).

To test for abnormalities in Sevenless signaling (17), we examined the retinas of mutant animals. Although reduced in number, the ommatidia of NF1P2 homozygote, ofNF1P2/NF1P1 , and ofNF1P2/Df(3R)boss15 flies (18) had the normal complement of photoreceptor and accessory cells (Fig. 3D). In homozygotes ofNF1P1 , 25% of ommatidia had one or more extra photoreceptor cells (6). However, this phenotype may result from the deletion of genes within the neurogenic E(spl)complex. A particularly sensitive indicator of Sevenless pathway function is the sevE4 ;SosJC2/+ mutant combination (19). Flies of this genotype that were also heterozygous forNF1P2 had no alteration in the fraction of ommatidia that had R7 cells. Thus, at least two Ras1-mediated signaling pathways downstream of RTKs are not influenced by a reduction inNF1 function.

Heterozygous loss of Ras1 or Sos (20) had no effect on the size of NF1 mutant pupae, nor did expression of an activated Rafgof mutation (Table 1). Thus, neither reducing nor increasing signaling through the Ras1-Raf pathway modifies the NF1 phenotype. Heterozygous loss of another Drosophila Ras-GAP homolog, Gap1(21), did not enhance the phenotype of NF1mutants. However, homozygous loss of both Gap1 andNF1 resulted in lethality during larval development (6). Thus, similar to what has been found for murine p120GAP and NF1 (22), Drosophila Gap1 and NF1 may have partially redundant roles as Ras regulators.

Flies carrying a viable heteroallelic combination of mutant alleles of the gene encoding the PKA catalytic subunit, DC0, are reduced in size (23). To test whether the adenosine 3′,5′-monophosphate (cAMP)–PKA pathway might represent a target for NF1, we examined pupae of another heteroallelic combination,DC0TW2/DC0B3 , and found them to be phenotypically similar to NF1 mutants (6). We then tested whether increasing PKA activity in NF1 mutant animals would rescue the size defect. A constitutively active murinePKA* transgene was expressed in an NF1 mutant background (20). Heat shock–induced expression of this transgene resulted in lethality. However, we were able to achieve lower transgene expression by growing the cultures at 28°C. Under these conditions, statistically significant rescue of the pupal size defect was observed (Table 1). In contrast to its effect on NF1mutant pupae, the PKA* transgene did not modify the phenotype of Tubby, a mutation that results in pupae of small size (24). Because expression of activated PKA suppressed the phenotype of null alleles of NF1, PKA appears not to function upstream of NF1 in a simple linear pathway. Therefore, PKA must either function downstream of NF1 or in a parallel pathway.

NF1 mutants also differ from wild-type flies in an assay that determines the number of flies that fly away upon release from their containers, either spontaneously or after repeated prodding (25). About 15% of either NF1 mutant (n = 200) failed to respond, whereas only 3% of parental K33 flies did not respond. The reduced escape rate does not reflect obvious anatomical defects of the peripheral nervous system or the musculature, and the mutants scored within normal limits in tests measuring their response to visual or olfactory stimuli (26). Electrophysiological studies showed that the mutants have a defect at the larval neuromuscular junction that is rescued by pharmacological manipulation of the cAMP-PKA pathway and that is insensitive to manipulation of Ras1-mediated signaling (27). Thus at least two NF1-deficient phenotypes can be rescued by increasing cAMP-PKA signaling. Further analysis ofNF1-deficient flies may define how the NF1 protein mediates cross talk between the Ras and PKA pathways and may ultimately suggest new therapeutic strategies for the treatment of human neurofibromatosis type 1.

  • * Present address: Division of Pathology, The Hospital for Sick Children, 555 University Avenue, Toronto M5G 1X8, Canada.

  • To whom correspondence should be addressed. E-mail: abernards{at}


Stay Connected to Science

Navigate This Article