PAK-kinase: Biological Overview | Evolutionary Homologs | Regulation | Developmental Biology | Effects of Mutation | References

Gene name - PAK-kinase

Synonyms -

Cytological map position - 83E

Function - signaling

Keywords - dorsal closure, axon guidance, cytoskeleton

Symbol - Pak

FlyBase ID: FBgn0014001

Genetic map position - 3-

Classification - PAK family kinase

Cellular location - cytoplasmic



NCBI links: Precomputed BLAST | Entrez Gene | UniGene |
BIOLOGICAL OVERVIEW

Paks (p21-activated kinases) are evolutionarily conserved regulators of the actin cytoskeleton. Drosophila PAK kinase, initially isolated on the basis of its homology to other PAK proteins, interacts with the small GTP-binding proteins Rac and Cdc42 and functions in signal transduction during dorsal closure (Harden, 1996). The yeast Pak Ste20 regulates polarized cell growth in response to mating pheromones, and mammalian Paks reorganize the actin cytoskeleton when overexpressed in tissue culture cells. Pak consists of an N-terminal regulatory region that inhibits the activity of the C-terminal kinase domain. The regulatory region contains binding sites for at least three signaling proteins: an N-terminal proline-rich sequence (PXXP) binds to Nck; a CRIB (Cdc42/Rac interactive binding) motif binds to GTP-bound forms of Cdc42 and Rac, and a proline-rich motif is constitutively bound to Pix, a guanine nucleotide exchange factor specific to Rac and Cdc42. In the inactive state, an autoinhibitory sequence adjacent to the CRIB motif inhibits Pak kinase activity (Hing, 1999 and references).

These biochemical studies have provided a basis for a model of Pak function in growth cones. In this model, recruitment of the Pak/Pix complex to the membrane by mammalian Nck (Drosophila homolog Dreadlocks) in response to guidance receptor activation promotes GTP binding and activation of the p21 GTP binding proteins Cdc42 and Rac by Pix. Cdc42/Rac can, in turn, bind the CRIB motif of Pak and induce a conformational change in Pak. This displaces the autoinhibitory peptide, thereby activating Pak kinase, which also explains the name of this group: p21 activated kinases, or PAKs. Hence, recruitment of the Pak/Pix complex to the membrane may lead to the activation of both Rho family GTPases and Pak (Hing, 1999 and references).

Dreadlocks (Dock), an SH3/SH2 adaptor protein, links guidance signals to changes in the actin-based cytoskeleton in photoreceptor (R cell) growth cones (Garrity, 1996). The compound eye of the fly contains some 800 simple eyes called ommatidia. Each ommatidium contains eight R cells (R1-R8) that project in a retinotopic fashion to two different layers in the brain. R1-R6 terminate in the lamina, whereas R7 and R8 terminate in the medulla. While dock mutant R cells extend axons into the optic lobe normally, they form abnormal patterns of connections in both the lamina and medulla with defects in topographic map formation and ganglion target specificity (i.e., lamina versus medulla). By analogy to Grb2, an adaptor protein that couples receptor tyrosine kinases (RTK) to Ras, it is hypothesized that Dock links guidance receptors to the related Rho family GTPases, thereby modulating the actin cytoskeleton. While Drosophila guidance receptors linked to Dock have not been identified, the SH2 domain of Nck can bind to the cytoplasmic regions of several mammalian guidance receptors upon ligand stimulation, including c-Met (Kochhar, 1996) and EphB1 (Stein, 1998). In addition, mammalian Nck binds through its SH3 domains to proteins such as Pak and Prk2, which in turn bind to and are activated by Rho family GTPases. As a step toward understanding how Dock regulates downstream effectors, the function in R cell axon guidance of a Drosophila homolog of one of these, Pak (p21-activated kinase) was examined (Hing, 1999 and references).

Dock and Drosophila PAK-kinase have been shown to physically interact as do mammalian NIK and PAK (Bokoch, 1996 and Lu, 1997). If Dock and Pak function in the same signaling pathway in developing R cells to regulate growth cone motility, they would be expected to colocalize in these cells. R cells extend their axons into the optic lobe during the third instar of larval development. Dock staining is markedly enriched in the lamina and medulla neuropils, consistent with its localization to R cell axons and growth cones (Garrity, 1996). In contrast, Dock is only expressed at low levels in the cell bodies of developing R cells as well as in neuronal cell bodies in the cortical regions in the lamina and medulla. Strong staining is seen, however, in the neuropils of the optic disc. The lamina and medulla neuropils contain no neuronal cell bodies and comprise axonal processes and growth cones. Both anti-Dock and anti-Pak antibodies stain the medulla neuropil uniformly, indicating that these proteins are expressed on many visual system fibers. Since the R7 and R8 axons only contribute a small fraction of the total number of fibers in the medulla, it is not possible to assess whether Pak and Dock are coexpressed in these axons. In contrast, at this stage in development, the vast majority of the processes in the lamina neuropil belong to R cells, including the expanded R1-R6 growth cones and axons of R7 and R8. Hence, Pak, like Dock, preferentially localizes to axons and growth cones (Hing, 1999).

To assess whether Pak is required for growth cone guidance, mutations disrupting its function were identified. It was assumed that, like dock, null mutations in Pak would cause recessive lethality. Accordingly, lethal mutations were identified in a small region of the chromosome within which Pak maps. From 9440 mutant lines containing randomly mutagenized third chromosomes, 238 lethal mutations mapping to a deficiency that deletes Pak were isolated. These were then tested against the same deficiency chromosome bearing a Pak-containing cosmid. The cosmid rescued 21 mutations. These fell into two groups based on complementation tests. A Pak cDNA expressed under the control of the heat shock promoter rescued the lethality associated with one complementation group, indicating that these mutations disrupt Pak function. These alleles are designated Pak1 to Pak13 (Hing, 1999).

To assess R cell projections in Pak mutants, eye-brain complexes from transheterozygous larvae were stained with the R cell-specific antibody mAb24B10. In wild type, R cell axons grow from the eye disc, through the optic stalk, and into the optic ganglia during the third instar of larval development. The eight R cell axons from each ommatidium form a single bundle. These bundles spread out upon entering the optic lobe and form a smooth topographic map that reflects the arrangement of ommatidia in the eye. Growth cones of R1-R6 are seen as a band of immunoreactivity. In contrast, individual R8 growth cones are readily observed in the medulla neuropil. They are evenly spaced and exhibit a characteristic expanded morphology. At this stage of development, few of the R7 axons stain with mAb24B10 (Hing, 1999).

In Pak strong loss-of-function mutants, R cell axons extend into the brain normally. However, these fibers do not spread evenly within the lamina and medulla. As a result, some regions are hyperinnervated while others lack innervation. In the medulla neuropil, R cell axons fail to find their proper targets but instead, terminate as thick, blunt-ended fascicles. Hence, in contrast to wild type, Pak mutant R cells do not elaborate a smooth topographic map in the lamina and medulla neuropils. A small fraction of the R2-R5 neurons project through the lamina and into the medulla, indicating a modest disruption in ganglion target specificity. Eye-specific expression of a wild-type Pak cDNA under the control of the GMR promoter rescues the mutant phenotype. These data indicate that Pak is required for axon targeting but that it is not required for axon outgrowth because in these mutants R cell axons extend in the correct direction and into the target region. The Pak phenotypes are essentially indistinguishable from those previously described (Garrity, 1996) in dock mutants. While Pak has a profound effect on R cell projections, it does not disrupt R cell fate determination or differentiation (Hing, 1999).

R cell axons not only target to specific regions of the developing optic ganglia, they play an essential role in inducing optic ganglion development. They induce the proliferation of neuronal precursor cells in the lamina, and subsequently they induce neuronal differentiation. R cell axons also induce lamina glial cell differentiation. These inductive processes were assessed in Pak mutants using BrdU labeling to detect proliferating lamina precursor cells, and anti-Elav and anti-Repo staining to assess lamina neuron and glial cell differentiation, respectively. These steps occur normally in Pak mutants. Hence, like dock, Pak is not required for lamina induction (Hing, 1999).

Mutation analysis of Pak reveals that the kinase domain, the Crib domain (which interacts with Rac/Cdc4), and the Dock-binding site are all required for Pak function in R cell axon guidance. To assess whether Pak acts downstream of Dock, a dominant gain-of-function form of Pak was constructed. Previous studies have demonstrated that membrane localization of human Pak1 to the membrane leads to kinase activation. Accordingly, a membrane-anchored version of Pak was constructed by fusing a myristylation signal from Drosophila Src1 to the N terminus of Pak and expressed in developing R cells. This construct is designated GMR-Pakmyr. A single copy of GMR-Pakmyr rescues the Pak mutant phenotype, indicating that Pakmyr retains wild-type activity. GMR-Pakmyr shows a dose-dependent dominant phenotype in R cells. Wild-type larvae carrying a single copy of GMR-Pakmyr show mild clumping of axons in the lamina cortex. The R8 projections in the medulla and the pattern of R cells in the eye disc, however, are normal. This defect is slightly enhanced in animals with two copies of GMR-Pakmyr. In larvae carrying four copies of GMR-Pakmyr, the R cell projection pattern is severely disrupted. In addition, the pattern of R cell clusters in the eye disc is highly abnormal. Remarkably, in these animals R cells delaminate from the eye disc epithelium and migrate through the optic stalk into the brain. Both membrane localization and kinase activity are required to induce motility, since larvae carrying four copies of either GMR-Pakmyr K459A (kinase inactive) or GMR-Pakwt (not membrane tethered) are indistinguishable from wild type. While kinase activity is necessary to induce guidance defects and cell motility, it is not sufficient. The R cell projection pattern in larvae carrying four copies of GMR-PakL115F, which encodes a constitutively active but soluble form of Pak, was also indistinguishable from wild type. In summary, the ability of GMR-Pakmyr to rescue a Pak mutant phenotype and to confer a dose-dependent dominant phenotype is consistent with it acting as a dominant gain-of-function allele (Hing, 1999).

If Pak functions downstream of Dock, then the constitutively active form of Pak should rescue Dock mutations. If a key step in Dock function is to recruit Pak to the membrane, then membrane-tethered Pak, GMR-Pakmyr, may rescue some aspects of the dock mutant phenotype. To test this, a single copy of the transgene was introduced into a dock null background. In dock mutants, R cell axons form large abnormal fascicles in the optic ganglia. This leads to hyperinnervated regions separated by areas lacking innervation in both the lamina and medulla. In addition to disrupting targeting, R cell axon terminals in the medulla are thick and blunt ended. The dock phenotype was shown to be substantially rescued by GMR-Pakmyr. Axon bundles between the lamina and medulla are thinner in rescued flies. Growth cones in the medulla are expanded and spread out more evenly, giving rise to an array of terminals. Quantification of the expanded growth cones in the medulla shows an increase from less than 2 in dock mutants to 64 in dock mutants carrying a copy of GMR-Pakmyr. This represents a restoration of about half the number of growth cones, compared with wild-type preparations of similar age. Rescue requires both myristylation and kinase activity; rescue is not observed with GMR-Pakwt, kinase inactive or constitutively active soluble Pak. These data are consistent with models in which recruitment of Pak to the membrane by Dock is an essential regulatory step in R cell axon guidance (Hing, 1999).

While these studies demonstrate that Pak is a primary target of Dock in R cell growth cones, little is known about the upstream signals that regulate Dock and Pak function. It is proposed that guidance signals, functioning through membrane bound receptors, recruit Dock and Pak to the membrane. Two observations support this view: (1) overexpression of a constitutively membrane-tethered form of Pak (Pakmyr) leads to marked changes in growth cone and cell motility, and (2) Pakmyr expressed at a lower level substantially rescues the Dock null mutant phenotype. The latter result is somewhat surprising, since it was envisioned that precise spatial regulation of Pak within the membrane by recruitment to activated guidance receptors is crucial to its guidance function, an activity that is unlikely to be replicated entirely by the myristylation tag. Activated guidance receptors may, in addition to recruiting Dock, lead to increased levels of phosphoinositides, which in turn provide an alternate route to recruit Pak/Pix into complexes induced by the activated guidance receptors. Alternatively, guidance signals may independently activate Cdc42 (or Rac) in specific spatial domains within the growth cone. These proteins may subsequently recruit Pakmyr to these regions through the CRIB domain. Although it is envisioned that recruitment of Pak to specific spatial domains in growth cone membranes is important in axon guidance, the possibility that general localization of Pak to the membrane is sufficient to promote its function in guidance cannot be ruled out (Hing, 1999).

The importance of regulating Pak activity in cells is underscored by the observations that misexpression of high levels of Pakmyr, beyond that necessary to rescue the dock mutant phenotype, induces these cells to delaminate from the eye disc epithelium and migrate into the brain. The recruitment of Pak and Dock to the membrane may nucleate formation of large signaling complexes. In addition to the potential link of Pak to tyrosine kinase signaling pathways through Dock, Pak also contains an extreme C-terminal domain that binds to the Gß subunit of trimeric G proteins. Indeed, the yeast Pak Ste20 is activated through this site in response to G protein signaling pathways stimulated by mating pheromone (Leeuw, 1998). Pix not only contains a Pak interaction site and a guanine nucleotide exchange factor activity, it also contains a PH domain that potentially links Pak to yet additional pathways, including those regulated by phosphoinositides. And finally, Dock also interacts with multiple proteins, including phosphatases (for example, Drosophila PTP61F; Clemens, 1996) and kinases, as well as other cytoskeletal regulators (e.g., N-WASP; McCarty, 1998). While a Dock/Pak complex is well suited to integrate multiple signals and transmit them to the changes in the actin-based cytoskeleton, Dock and Pak also may act separately in different combinations to regulate growth cone guidance (Hing, 1999).


GENE STRUCTURE

cDNA clone length - 3163

Bases in 5' UTR - 268

Bases in 3' UTR - 783


PROTEIN STRUCTURE

Amino Acids - 704

Structural Domains

Two blocks of highly conserved sequences have been identified in PAK-kinase of Drosophila: the serine/threonine kinase domain and a stretch of amino acids identified as the p21-binding domain. The p21-binding domain serves to bind to small GTP-binding proteins such as Rac and Cdc42. The p21-binding domain is most homologous to that of the original rat PAK sequence and to domains from C. elegans and human PAK-Ste20 family members with homologies on the order of 80%. There is considerably less homology to the corresponding domains of yeast Ste20 (Harden, 1996).


PAK-kinase: Evolutionary Homologs | Regulation | Developmental Biology | Effects of Mutation | References

date revised: 30 July 99

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.