BIOLOGICAL OVERVIEW

pelle mutations cause a dorsalized embryo (one lacking ventral structures) whose phenotype can be corrected by injecting mutant eggs with the cytoplasm from normal eggs. The mRNA for Pelle is provided maternally; therefore mutant mothers are unable to support the growth of embryos. Normal maternal cytoplasm or mRNA when injected into embryos from mutant mothers partially restores the normal pattern. Rescuing potential remains in the cytoplasm of normal eggs until the late blastoderm stage. After this stage has passed, eggs from mutant mothers fail to respond to injected mRNA. Thus there is a critical period in development in which the pelle gene product can act to rescue defective eggs (Müller-Holtcamp, 1985).

The cloning of pelle kinase has confirmed classic injection experiments. Kinases are proteins that can catalytically attach phosphate groups, thus passing a signal from one molecule (the kinase) to its target. The message that Pelle transduces, the activation of the Toll receptor by its ligand Spätzle, is passed to Cactus causing its degradation. Freeing the Rel-related protein Dorsal from it cytoplasmic inhibitor Cactus, allows Dorsal to translocate into ventral and ventrolateral nuclei and direct gene expression.

Pelle-mediated signaling induces the spatially graded degradation of Cactus. Using a tissue culture system which reconstitutes Pelle-dependent Cactus degradation, it can be shown that a motif in Cactus, resembling the sites of signal-dependent phosphorylation in the vertebrate Cactus homolog IkappaB is essential for Pelle-induced Cactus degradation. Substitution of four serines within this motif with nonphosphorylatable alanine residues generates a mutant Cactus that still functions as a Dorsal inhibitor but is resistant to degradation (Reach, 1996).

Although both Cactus and Dorsal are modified by phosphorylation in response to signaling, it is believed that signal-dependent modification of Cactus precedes modification of Dorsal. Signal-dependent modification of Dorsal presumably occurs downstream of Cactus degradation, since loss of Cactus function induces Dorsal modification in the absence of signaling. Phosphorylation of Dorsal upon release from Cactus could reflect unmasking of a phosphorylation site or an encounter with a protein kinase upon translocation into nuclei (Gillespie, 1994 and Reach, 1996).

The Pelle kinase appears to require Tube for its activation. Tube protein associates with the membrane of the oocyte and is required to pass the activation signal from the receptor Toll to Pelle. In the process Tube also becomes phosphorylated by Pelle. Whether Pelle can phosphorylate Cactus or Dorsal is unknown. Tube is the one known target for Pelle phosphorylation (Grosshans, 1994 and Galindo, 1995).

Biochemical interactions are described among recombinant Toll, Pelle and Tube that provide insights into early events in activation of the signaling cascade. A tertiary complex exists prior to activation. The Pelle-Toll complex is required to suppress that kinase activity of Pelle. Pelle binds directly to a region within the Toll intracytoplasmic domain, providing the first evidence that these two evolutionarily conserved molecules physically interact. Pelle contains an N-terminal putative regulatory domain, consisting largely of a region with significant similarity to the consensus death domain, and a C-terminal catalytic domain. To determine whether either (or both) of these regions are required for the interaction with the intracellular domain of Toll (Toll IC), the Pelle N terminus (Pelle-R) and C terminus (Pelle-C) were produced separately by in vitro translation, and then tests were carried out to see if these molecules could bind any of several GST-Toll IC derivatives. The results indicate that the Pelle N terminus is incapable of binding any of the GST-Toll derivatives, whereas the catalytic region binds efficiently to both GST-Toll and GST-Toll ICNae (a derivative lacking the C-terminal ID but containing the entire IL-1R homology region). The Toll C-terminal inhibitory domain (ID) is neither necessary nor sufficient for this interaction (Shen, 1998).

It is thought that upon activation Pelle is autophosphorylated, and that this prevents binding to Toll as well as Tube. Autophosphorylation occurs in the N-terminal, death-domain-containing region of Pelle, a region dispensable for binding to Toll but required for enzymatic activity. Pelle phosphorylates Toll, within the region required for Pelle interaction, but this phosphorylation can be blocked by a previously characterized inhibitory domain at the Toll C terminus. These and other results allow for the proposal of a model by which multiple phosphorylation-regulated interactions between these three proteins lead to activation of the Dorsal signaling pathway (Shen, 1998).

It is proposed that the intracytoplasmic IL-1R homology domain of the Toll receptor initially interacts directly with unphosphorylated Pelle, and that the Toll ID helps down regulate kinase activity. Tube is also recruited to the complex through its interaction with Pelle and/or Toll, forming the Toll/Pelle/Tube ternary complex (Pelle inactive). Binding of Sp”tzle to Toll induces Toll dimerization, and the Toll cytoplasmic domain is modified through conformational changes or proteolysis of the Toll inhibitory domain, allowing activation of kinase activity (Pelle active). Active Pelle then phosphorylates multiple substrates, including itself, Toll and Tube. This causes disruption of the Toll/Pelle/Tube complex, freeing Pelle to phosphorylate unknown downstream targets, eventually resulting in Cactus phosphorylation/degradation and Dorsal phosphorylation (e. g., by protein kinase A [PKA] and possibly Pelle itself] and nuclear translocation. Phosphorylated Tube may also translocate with Dorsal and function as a transcriptional coactivator (Shen, 1998).

The direct interactions described between unphosphorylated Pelle, Toll and Tube are consistent with the existence of a ternary complex at the plasma membrane. Pelle interacts with Toll via residues in its catalytic domain, and with Tube via Pelle's N-terminal death domain: both interactions can occur simultaneously. An important question is whether the ternary complex forms independent of signaling. Previous studies have shown that the artificial recruitment of Pelle or Tube to the plasma membrane can initiate the signaling pathway independent of ligand binding. But it is not clear whether it is recruitment to the membrane per se that results in activation, or the dimerization of the Torso fusion proteins employed in these previous studies. A possible mechanism for Pelle activation is simply dimerization, induced naturally, it is suggested, by conformational changes in the ternary complex that occur following ligand binding. How might such changes be induced? There is considerable indirect evidence suggesting that Toll molecules interact: an attractive model posits that ligand binding induces dimerization or even aggregation. It is suggested that this leads to activation of signaling, i.e., of Pelle activity, by either (or both) of two related mechanisms: (1) oligomerization of Toll receptors increases the local ternary complex concentration and hence Pelle concentration, thereby favoring Pelle dimerization and activation by simple mass action; (2) ligand-induced Toll self-association causes a conformational change in the intracytoplasmic domain such that the ID is displaced, thereby facilitating Pelle activation, again perhaps by dimerization. A speculative possibility is that the ID is actually cleaved upon activation. The product of the strong dominant gain-of-function allele Toll 10b, which contains a single C to Y change in its extracellular domain, has been found in a partially proteolyzed form, such that full-length Toll 10b is associated with a truncated form lacking most or all of its extracellular domain as well as likely sequences from the very C terminus, i.e., the ID. Perhaps relevant to this, a putative PEST degradation sequence is situated between the IL-1R homology region and the ID. It is intriguing that the structure of this truncated product is similar to mammalian IL-1RAcp, which functions in IRAK activation during IL-1 signaling. In any event, it is proposed that IL-1R homology domain interactions activate Pelle via the direct, phosphorylation-sensitive protein-protein interactions described in this paper (Shen, 1998).

GENE STRUCTURE

Bases in 5' UTR - 172

Bases in 3' UTR - 202

PROTEIN STRUCTURE

Amino Acids - 501

Structural Domains

DNA sequence analysis revealed that pelle encodes a protein of 501 amino acids, the last 292 comprising a protein kinase catalytic domain. Microinjection of in vitro synthesized transcripts containing site-directed mutations indicates that the kinase catalytic domain is required for biological activity. This domain is most similar to that of the RAB and MOS protein kinases and is predicted to have a serine and threonine specificity. The identity between Pelle and Drosophila Raf, a kinase acting in the ras pathway, is equivalent to that between Pelle and human Raf (27% and 26.6% respectively) (Shelton, 1993).

The interaction between the death domains (DDs) of Tube and the protein kinase Pelle is an important component of the Toll pathway. Published crystallographic data suggests that the Pelle-Tube DD interface is plastic and implies that in addition to the two predominant Pelle-Tube interfaces, a third interaction is possible. The NMR solution structure of the isolated death domain of Pelle is presented along with a study of the interaction between the DDs of Pelle and Tube. The data suggests the solution structure of the isolated Pelle DD is similar to that of Pelle DD in complex with Tube. Additionally, they suggest that the plasticity observed in the crystal structure may not be relevant in the functioning death domain complex (Moncrieffe, 2005).

The crystal structure suggests that the Pelle­Tube DD complex can exist as a tetramer comprised of Pelle and Tube heterodimers arranged in a linear sequence P1:T1:P2:T2 and whether this persists in solution needs to be determined as it may have implications for the function of the DD complex. To address these issues, the structure of the isolated Pelle-DD in solution and the interaction between the DDs of Pelle and Tube have been solved using nuclear magnetic resonance (NMR) spectroscopy. The results suggest that in the beta-helical regions, the structure of Pelle-DD is similar to that of Pelle-DD in complex with the DD of Tube, and of the two types of Pelle-Tube dimer interfaces observed in the crystal structure (P1:T1 and P2:T2) only one is likely to persist in solution (Moncrieffe, 2005).

The NMR data presented confirm the extensive nature of the Pelle-Tube death domain interface because all twenty seven residues that constitute the core interaction surface on Pelle-DD show significant chemical shift differences between the ¹H-¹⁵N HSQC spectrum of Pelle-DD and the ¹H-¹⁵N HSQC-TROSY spectrum of the Pelle-Tube DD complex. The NMR data also suggest that the Tube1:Pelle2 interaction observed in the crystals, may persist in solution. This implies that the death domains of Pelle and Tube are capable of forming a tetramer and this is corroborated by the sedimentation velocity data which reveals that the concentration of the tetrameric complex is very small (0.4%) relative to that of the dimer. The dominant species formed by the interaction of Pelle-DD and Tube-DD is a dimer and whether the very small amount of the tetrameric species plays a role in signaling is unknown; for example, a mutant of Pelle bearing a mutation in the Tube1:Pelle2 interface (D50K) fails to express protein. It is conceivable that the residues in Pelle that are part of the Tube1:Pelle2 interface are sites of interaction for other members of the signaling complex. A recent report argues against that partner being dMyD88 since this component is thought to bind predominantly to Tube. A more likely scenario however, given that dMyD88 appears to bind predominantly to Tube and not Pelle, is that in the functioning complex, P2 is occupied by dMyD88. Thus, the residues in Tube-DD that contribute to the T1:P2 interface are likely to be the scaffold that recruits dMyD88 (Moncrieffe, 2005).

In summary, the structure of the isolated Pelle DD is similar to that of the Pelle DD in complex with the DD of Tube. While the death domains of Pelle and Tube may form a tetramer, consistent with the crystallographic observation the concentration of this species is extremely low. Consequently, inferences regarding the “plasticity” of the Pelle-Tube DD complex in vivo should be reevaluated (Moncrieffe, 2005).

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.