Requirement of Drosophila NF1 for Activation of Adenylyl Cyclase by PACAP38-Like Neuropeptides

See allHide authors and affiliations

Science  02 May 1997:
Vol. 276, Issue 5313, pp. 795-798
DOI: 10.1126/science.276.5313.795


The human neurofibromatosis type 1 (NF1) tumor suppressor protein functions as a Ras-specific guanosine triphosphatase–activating protein, but the identity of Ras- mediated pathways modulated by NF1 remains unknown. A study of Drosophila NF1 mutants revealed that NF1 is essential for the cellular response to the neuropeptide PACAP38 (pituitary adenylyl cyclase–activating polypeptide) at the neuromuscular junction. The peptide induced a 100-fold enhancement of potassium currents by activating the Ras-Raf and adenylyl cyclase–adenosine 3’,5’-monophosphate (cAMP) pathways. This response was eliminated in NF1 mutants. NF1 appears to regulate the rutabaga-encoded adenylyl cyclase rather than the Ras-Raf pathway. Moreover, the NF1 defect was rescued by the exposure of cells to pharmacological treatment that increased concentrations of cAMP.

Mutations in the human NF1 gene lead to a common genetic disorder that is identified by benign tumors of the peripheral nerves, hyperpigmentation, white matter lesions in the brain, learning disabilities, and many other manifestations (1, 2). The NF1 protein, which contains a fragment similar to the guanosine triphosphatase (GTPase)–activating protein for Ras (Ras-GAP), stimulates the intrinsic activity of Ras-GTPase and therefore inhibits biological activation of Ras (3). However, NF1 may not act solely to regulate Ras but may also function as an effector that mediates signaling important for differentiation [for a review, see (1)]. Our study of Drosophila NF1 mutants indicates that the activation ofrutabaga (rut)-encoded adenylyl cyclase (4,5) through heterotrimeric guanine nucleotide– binding protein (G protein)–coupled receptors is regulated by NF1.

The Drosophila homolog of NF1 is 60% identical to the human NF1 protein, neurofibromin, over its entire 2802–amino acid length (6). Although homozygous loss of the Nf1gene in mice is lethal (7), two viable Drosophilanull mutations of NF1 (NF1 P1 andNF1 P2) have been generated. No NF1 was detected by protein immunoblotting in these two mutants (6).NF1 P1 is a small deletion that includes theNF1 locus and at least two adjacent genes, andNF1 P2 is a P-element insertion (6). Modulation of voltage-activated K+ currents (8, 9) induced by the neuropeptide pituitary adenylyl cyclase–activating polypeptide (PACAP38) is eliminated in these two mutant alleles.

PACAP38 [which belongs to the vasoactive intestinal polypeptide-secretin-glucagon peptide family and stimulates cAMP synthesis through G protein–coupled receptors in vertebrates (10)] induces a 100-fold enhancement of K+currents by coactivating both Rut–adenylyl cyclase–cAMP and Ras-Raf kinase pathways (9). Mutations in the rut(4), Ras (11), or raf(12) loci eliminate the response to PACAP38 (9). Activation of both cAMP and Ras-Raf pathways together, but not alone, mimics the PACAP38 response (9). The involvement of Ras in the PACAP38 response led us to investigate the effect of NF1mutations in Drosophila.

PACAP38-induced responses were recorded by the two-microelectrode voltage-clamp method from body wall muscle fibers of larvae at the third instar (9, 10, 13, 14). Perfusion of PACAP38 to the neuromuscular junction induced an inward current followed by a 100-fold enhancement of K+ currents in wild-type larvae (8,9) (Fig. 1, A and B). InNF1 P1 and NF1 P2 mutants, the inward current remained mostly intact (Fig. 1A), but the enhancement of K+ currents was abolished (Fig. 1, B and C). Because the inward current is not affected in NF1 mutants, it appears that PACAP38 receptors are normally activated by the peptide in these mutants.

Figure 1

Elimination of PACAP38-induced enhancement of K+ currents in NF1 mutants. (A) PACAP38-induced synaptic inward current inNF1 P1 and NF1 P2mutants. Representative examples of inward current traces induced by 5 μM PACAP38 are shown. The horizontal bar above the traces indicates the period during which PACAP38 was perfused. The histogram compares the average peak amplitude ± SEM of the inward current among different genotypes. The number of muscle fibers recorded is shown over the error bar. WT, wild type. (B) Failure of PACAP38 to enhance voltage-activated K+ currents inNF1 mutants. K33 is the parental line from whichNF1 mutants were generated (6). The arrow indicates focal application of PACAP38 (5 μM). In all figures, the time in minutes after pressure-ejection of PACAP38 is indicated at the top of the current traces. The number of muscle fibers/the number of larvae recorded (N/n) = 31/21, 4/3, 8/5, and 25/18 for wild-type, K33, NF1 P1, andNF1 P2, respectively. (C) Current-voltage (I-V) relations before or 2 min after application of PACAP38 (5 μM). All wild-type and K33 but none of mutant muscle fibers responded. Data in this and all other figures (except Fig. 2B) were recorded with extracellular saline containing 1 mM Ca2+.

To rule out potential developmental effects of the NF1mutation, we studied transgenic flies carrying an inducible normalNF1 gene. The hsNF1 transgene was expressed after heat shock in the transgenic NF1 mutants hsNF1;NF1 P1and hsNF1;NF1P2. PACAP38-induced enhancement of K+ currents was observed in hsNF1;NF1 P1larvae subjected to heat shock (37°C for 1 hour) but not in those without heat shock (Fig. 2A). The hsNF1; NF1 P2 larvae, however, showed a normal response to PACAP38 even in larvae not subjected to heat shock. This was probably the result of constitutive expression of the hsNF1 transgene because a large amount of NF1 protein was detected in these flies (6). To reduce the amount ofhsNF1expression, we selected hsNF1;NF1P2/+; NF1 P2larvae in which only one copy of the hsNF1transgene was present. In these larvae, the PACAP38 response was only observed after heat shock (Fig. 2A). The PACAP38-induced enhancement was fully rescued 4 hours after heat shock but was observed with a smaller enhancement as early as 1.5 hours after heat shock. Such a time course suggests that all other components in the PACAP38 signaling pathways remain intact so that the preparation resumes PACAP38 responsiveness as soon as enough NF1 is synthesized.

Figure 2

Rescue of NF1 mutant phenotype by induced expression of the hsNF1 transgene. (A) PACAP38-induced response. NF1 mutants were combined with the hsNF1 transgene controlled by a heat-shock promoter as hsNF1; NF1 P1 andhsNF1; NF1 P2. These larvae were subjected to heat shock at 37°C for 1 hour. Recording was done at various times after heat-shock induction. All traces were recorded about 4 to 6 hours after the heat shock. The PACAP38 response was observed as early as 1.5 hours after heat shock but with a smaller amplitude of enhancement. N/n = 8/6, 4/3, 4/3, 5/3, and 4/3 for panels from top to bottom, respectively. (B) Electrically evoked endogenous PACAP38-like enhancement of K+ currents rescued by induced expression of thehsNF1 transgene. Stimulation at 40-Hz of wild-type motor axons for 3 s induced a PACAP38-like enhancement of K+currents in 2 mM Ca2+ (9). Control currents were recorded before stimulation. During stimulation, muscle fibers were clamped to −80 mV. Evoked excitatory junctional current traces are shown under “40 Hz stimulation.” Then, the same voltage paradigm for eliciting K+ currents as that used in the control was repeated at various times (indicated at the top of the current traces) after stimulation. Scale bar, horizontal/vertical: 5 s/50 nA.

Because PACAP38 is a vertebrate peptide (10), we tested the response induced by endogenous PACAP38-like neuropeptide (8,9). High-frequency stimulation (40 Hz) applied to motor axons through a suction pipette increased K+ currents, presumably by causing the release of PACAP38-like peptides (8). This evoked PACAP38-like response was also eliminated in NF1mutants and rescued by the expression of the hsNF1transgene (Fig. 2B).

Because NF1 acts as a Ras-GAP (3), we examined two null alleles of the Drosophila gene Gap1:rI 533B1 and rI 533PB. Flies carrying the mutations have disrupted eye development that results from increased Ras activity (15). PACAP38 induced a normal enhancement of K+ currents (Fig. 3A) in both Gap1 mutants. Moreover, recordings from transgenic larvae showed that induced expression of constitutively active Ras (RasV al 12) (16) or active Raf protein kinase (Rafgof) (17) neither blocked nor mimicked the PACAP38 response (Fig. 3B) (9). These results suggest that failure to negatively regulate Ras-Raf signaling does not explain the defective PACAP38 response in NF1 mutants.

Figure 3

Normal PACAP38 response in GAP1 mutants and in transgenic larvae expressing active Ras. (A) Normal PACAP38 response in mutants carrying Gap1 null alleles rI 533PB andrl 533B1. N/n = 6/4 and 4/3, respectively. (B) Normal PACAP38 response in transgenic larvae expressing inducible active Ras (RasV al 12) after heat-shock induction (37°C, 1 hour).N/n = 7/5 and 3/2 for heat-shockedRasV al 12and wild-type larvae, respectively.

Application of the membrane-permeable cAMP analogs dibutyryl cAMP or 8-bromo-cAMP to the larval neuromuscular preparation was insufficient to produce the PACAP38-like enhancement of K+ currents (9, 18) and appeared not to disrupt the PACAP38 response in wild-type larvae (Fig. 4A). This implies that cAMP may not cause inhibition of Raf activity as reported in other preparations (19). Application of these cAMP analogs to NF1mutants did restore the normal response to PACAP38 (Fig. 4A). BothNF1 P2 homozygotes and heteroallelicNF1P1/NF1 P2 larvae showed enhanced K+ currents. NF1 P1 larvae also responded, but with a smaller amplitude of response (Fig. 4A), which may be a nonspecific effect of their genetic background because the response ofNF1P1/NF1 P2 heterozygotes to PACAP38 was fully restored by treatment with cAMP analogs.

Figure 4

PACAP38-induced response inNF1 mutants treated with cAMP analogs. (A) PACAP38 response in NF1 mutants treated with cAMP analogs. Dibutyryl cAMP (d. cAMP) (1 mM) was added before perfusion of PACAP38.N/n = 5/2, 19/14, 16/9, and 13/9 for wild-type, NF1 P2,NF1 P1, andNF1P1/NF1 P2, respectively. (B) Time course of the cAMP effect. Dibutyryl cAMP was applied 10 min before or 30 s, 2 min, or 2.5 min after the perfusion of PACAP38. (C) PACAP38 response inrut 1 mutants treated with cAMP analogs.N/n = 22/15 and 5/4 for without and with the cAMP analog, respectively. (D) Failure of the cAMP analogs to restore PACAP38 response in Ras mutants. Flies with Ras 12a, a weak allele, were viable but showed no PACAP38-induced enhancement of K+ currents (top panel) (9). N/n = 9/5 and 5/4 for without and with the cAMP analog, respectively.

The cAMP analogs were effective if applied any time before or within 2 min after application of PACAP38 (examples of 10 min before and 30 s and 2 min after are shown in Fig. 4B). After 2 min, cAMP analogs failed to enhance the response of NF1 P2mutants to PACAP38. This time course is consistent with a model whereby in NF1 mutants the Ras-Raf pathway is normally activated in response to PACAP38 for 2 min, but the cAMP pathway is blocked. Therefore, synergistic modulation of K+ currents can be achieved if cAMP analogs are supplied during the transient activation of the Ras-Raf pathway. Addition of cAMP analogs also restored the response to PACAP38 in rut 1 mutants, but not in Ras 12a mutants (Fig. 4, C and D) (4, 5, 9).

To further test whether activation of cAMP signaling rescues the defective PACAP response of NF1 mutants, we applied the drug forskolin, which stimulates G protein–coupled adenylyl cyclase activity (20, 21), to the neuromuscular preparation. PACAP38 induced a normal response in NF1 P2 andNF1P1/NF1 P2 mutants exposed to forskolin (Fig. 5A). This indicates that adenylyl cyclase is present but is not activated by receptors for PACAP38-like neuropeptides. Forskolin also restored the PACAP38 response in rut 1 mutants (Fig. 5B) even though the Rut–adenylyl cyclase is completely nonfunctional (4,5). It is possible that cAMP synthesized by other adenylyl cyclases upon forskolin stimulation is sufficient to modulate K+ currents together with the Ras pathway activated by PACAP38 (22).

Figure 5

PACAP38 response in NF1mutants treated with forskolin. (A) Effect of forskolin (fors.) in NF1 mutants. Forskolin (10 μM) was applied to the extracellular solution 4 min before perfusion of PACAP38 to wild-type or mutant larvae as indicated. Forskolin at 2 μM was also effective. N/n = 5/4, 21/16, 10/4, and 9/4 for wild-type, NF1 P2,NF1 P1, andNF1P1/NF1 P2 larvae, respectively. (B) Effect of forskolin inrut 1 mutants. Forskolin (10 μM) was applied 4 min before PACAP38. N/n = 5/4. (C) Adenylyl cyclase activity (26) in membrane fractions. The Ca2+ dependency of adenylyl cyclase assayed from the membrane fraction of fly abdomens is normal inNF1 P2 mutants. (D) Basal and forskolin-stimulated activities of adenylyl cyclase are similar in wild-type and NF1 P2 flies. Forskolin (100 μM) was added to the extracted membrane fraction.

Adenylyl cyclase shows abnormal subcellular localization in yeastIRA (inhibitor of Ras activity) mutants (23). TheIRA gene encodes proteins that are distantly related to NF1 and that are involved in mediating Ras-dependent activation of adenylyl cyclase. Although the yeast cyclase is very different from Rut–adenylyl cyclase and other cyclases in higher organisms (24,25), we examined adenylyl cyclase activity in membrane fractions (26). The Rut–adenylyl cyclase is the only cyclase that can be activated by Ca2+-calmodulin in the tissues from fly abdomen, as indicated by the lack of the Ca2+-dependent cyclase activity in rut 1 mutants (4,5) (Fig. 5C). In addition, the basal activity (4) and the forskolin-stimulated (21) activity of adenylyl cyclase were also reduced in rut 1 mutants. However,NF1 mutations did not affect the basal activity, the Ca2+-dependent activity, or the forskolin-stimulated activity of adenylyl cyclase (Fig. 5, C and D). Therefore, Rut–adenylyl cyclase is present in these membranes and can be normally activated by Ca2+-calmodulin and forskolin.

In summary, signaling by the PACAP38 neuropeptide is impaired inNF1 mutants, and the defect is apparently caused by a blockade of PACAP38-stimulated activation of Rut–adenylyl cyclase. Thus, the NF1 protein not only acts as a negative regulator of Ras but also as a crucial component for activation of the cAMP pathway. The induced expression of a catalytic subunit of cAMP-dependent protein kinase rescues the developmental phenotype of small body size inNF1 P1 and NF1 P2 mutants (6), providing further support for the above conclusion. Exploration of the mechanism by which NF1 influences G protein–mediated activation of adenylyl cyclase may lead to new insights into mechanisms of G protein– mediated signal transduction and the pathogenesis, and possibly the treatment, of human type 1 neurofibromatosis.

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


View Abstract

Navigate This Article