| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Biological Sciences, University of Windsor, Windsor, Ontario, Canada
1Correspondence: Department of Biological Sciences, University of Windsor, Windsor, Ontario, Canada N9B 3P4. E-mail: ahubber{at}uwindsor.ca
 |
ABSTRACT |
|---|
 TOP ABSTRACT BACKGROUNDSTRUCTURAL AND FUNCTIONAL...EVOLUTION OF CAP PROTEINSADDITIONAL FEATURES OF CAP...CONCLUSIONS AND PERSPECTIVES FOR...REFERENCES |
|---|
Key Words: CAPs • actin cytoskeleton • cell signaling • endocytosis • development
|
BACKGROUND |
|---|
|
TOPABSTRACTBACKGROUNDSTRUCTURAL AND FUNCTIONAL...EVOLUTION OF CAP PROTEINSADDITIONAL FEATURES OF CAP...CONCLUSIONS AND PERSPECTIVES FOR...REFERENCES |
|---|
, 2
). The majority of actin binding proteins interact with F-actin and serve to regulate assembly of the actin filaments and to link other proteins to the F-actin network. For example, capping proteins (which regulate the accessibility of barbed, fast-growing ends) and severing proteins (which produce new barbed ends) both require F-actin binding proteins (for recent reviews, see refs 3
, 4
). In contrast, G-actin binding proteins are less common and are believed to control the availability of unpolymerized actin in a sequestered form. Actin sequestering factors can either enhance polymerization or cause disassembly of F-actin. Examples of G-actin binding proteins include profilin (for a review, see ref 5
), ß-thymosins (see ref 6
), Wiskott-Aldrich syndrome protein (WASp) (see ref 7
for a review), and CAPs.
Monomeric actin polymerizes in a head-to-tail fashion to form a helical F-actin filament with a defined structural polarity (8)
. The reaction is dependent on ATP hydrolysis and this energy input gives the actin filament the ability to grow at one end while simultaneously shrinking from the other end (8)
. The cycling between actin polymerization and depolymerization is influenced by the concentration of G-actin in the cell. Cells contain large G-actin pools on the order of 50–200 µM. Paradoxically, once the concentration of G-actin rises above 0.1 µM, polymerization into F-actin occurs and proceeds until the G-actin concentration once again reaches 0.1 µM. This level is referred to as the critical concentration. Therefore, to account for the large difference between the amount of G-actin in cellular pools and that required for the critical concentration, it has been proposed that the majority of G-actin is bound to other proteins that maintain the critical concentration (9)
. The accessibility of these proteins to G-actin is highly regulated, although the triggers for release of G-actin from the sequestered actin pool during assembly of actin filaments are generally unknown.
A critical question is how cells translate extracellular or intracellular signals to changes in the cytoskeleton. Many exogenous factors activate signaling pathways that elicit changes in the cytoskeleton to control cell migration, adhesion, invasion, and cytokinesis. Although key regulators of these pathways have been characterized (e.g., Rac, Rho, and Cdc42; for reviews, see refs 10
11
12
13
), the downstream effectors in these pathways remain unclear. Moreover, the regulation of these pathways is specific for different cell types. Identification of the proteins involved directly or indirectly in linking signals generated from specific pathways to elements of the cytoskeleton is just beginning.
One class of potential bridging proteins is represented by the family of adenylyl cyclase-associated proteins (CAPs) first identified in yeast and subsequently isolated in all eukaryotic organisms studied (Table 1
). The first CAP gene (also called SRV2) was isolated in Saccharomyces cerevisiae as a suppressor of the activated RAS2Val19 allele (14)
. CAP was simultaneously isolated as a component of the adenylyl cyclase (Cyr1p) complex (15)
. S. cerevisiae cap- cells exhibit four major phenotypes: an inability to grow on rich medium, temperature sensitivity on minimal medium, sensitivity to nitrogen deprivation, and changes in cell morphology. Cells deficient in CAP are rounder and larger than normal cells, suggesting that cytoskeletal structures are disrupted.
|
S. cerevisiae has two Ras genes, RAS1 and RAS2, which signal through Cyr1p and protein kinase A (PKA) to regulate cell growth. Ras2p regulates the actin cytoskeleton through a PKA-independent pathway (16)
. Since Cyr1p is the effector of RAS in S. cerevisiae and eliminating CAP in yeast suppressed activated RAS signaling, it was theorized that CAP may provide a link between nutritional response signaling and changes in the cytoskeleton. The conservation of CAP homologues in all species suggests that CAPs play a fundamental role in cell growth and cytoskeletal organization (Table 1)
.
Not all CAPs have been shown to physically interact with adenylyl cyclase or regulate its function; only those present in S. cerevisiae (CAP/SRV2) and Schizosaccharomyces pombe (cap) show a direct interaction (17
, 18)
. For clarity, however, all cyclase-associated protein homologues regardless of species origin or adenylyl cyclase binding will be referred to as CAP in this review. Another common confusion with the name CAP lies in the number of proteins that share this acronym, including the family of capping proteins (e.g., CAPZ) involved in binding barbed ends of actin filaments and preventing polymerization. Standard nomenclature dictates that CAP in S. cerevisiae be referred to as Cap1p. However, since Cap1p refers to the product of the S. cerevisiae 
subunit capping protein gene CAP1, we will refer to the yeast cyclase-associated protein simply as CAP. The Cbl-associated protein (CAP) involved in insulin signaling and actin regulation shares the same acronym (19
, 20)
. The CAP proteins discussed here show no homology with F-actin binding proteins and do not perform similar functions.
At least two different CAP genes, CAP1 and CAP2, share 64% amino acid identity in mammals (21
, 22)
. The existence of multiple CAP genes in nonmammalian species is not documented. The roles of mammalian CAP2 proteins have not been studied extensively and will be discussed only briefly. We will focus on the structure and function of the monomeric actin binding CAPs and their roles in actin reorganization and cell signaling. The potential roles of CAPs in development and vesicle trafficking will be discussed.
|
STRUCTURAL AND FUNCTIONAL DOMAINS OF CAP PROTEINS |
|---|
|
TOPABSTRACTBACKGROUNDSTRUCTURAL AND FUNCTIONAL...EVOLUTION OF CAP PROTEINSADDITIONAL FEATURES OF CAP...CONCLUSIONS AND PERSPECTIVES FOR...REFERENCES |
|---|
-helices with a region rich in ß-sheets located in the carboxyl terminus. The four major structural and functional domains of S. cerevisiae CAP are illustrated in Fig. 1
. The amino-terminal domain of yeast CAP interacts with adenylyl cyclase and is sufficient for cellular responses to activated RAS protein. The carboxyl-terminal domain is necessary for normal cellular morphology and the response to nutrient stress. This domain binds monomeric actin in all CAPs tested to date. A third, centrally located, proline-rich domain has been shown to interact with Src homology 3 (SH3) domains of specific proteins. Finally, a fourth domain is responsible for oligomerizing CAP to form multimeric complexes.
|
Adenylyl cyclase binding domain
Yeast CAPs (S. cerevisiae and S. pombe) are unique within the CAP family as they are the only proteins shown to interact directly with adenylyl cyclase. Cyr1p in S. cerevisiae is activated by the RAS proteins Ras1/2p. Perturbations in RAS function cause a failure to grow, and activated mutant alleles of RAS2 (e.g., RAS2Val-19) display phenotypes identical to those observed when cyclic AMP pathways are activated. These phenotypes include sensitivity to heat shock and nitrogen starvation. In S. cerevisiae, cells containing mutations in the CAP gene were found to be unresponsive to activated RAS2Val-19, suggesting that CAP plays a role in the RAS2 signal transduction pathway (14
, 15)
. Upon further analysis, the amino-terminal 168 amino acids of CAP were shown to be sufficient to restore heat shock sensitivity to RAS2Val-19 cells (18)
. Conversely, small amino-terminal deletions in CAP (i.e., amino acids 2–31) impaired the heat shock sensitivity of activated RAS strains (18)
, implying that a short 31 amino acid domain in the NH2 terminus is essential for Cyr1p activation. Mutagenesis studies revealed that the carboxyl-terminal 150 amino acids of yeast Cyr1p interact with the amino-terminal domain of CAP. This interaction is sufficient to suppress the heat shock sensitivity of RAS2Val-19 cells (23)
. The amino-terminal 36 residues of CAP were sufficient to interact with the carboxyl-terminal 119 amino acids of CYR1 (21)
. Using in vitro reconstitution cyclase activation studies (24)
, the amino-terminal 168 residues of CAP were shown to be required for Cyr1p activation by activated Ras2p, which matched the CAP domain required for the in vivo cAMP response (18)
. Therefore, although a large amino-terminal region of CAP is essential for cyclase activation in vitro, only the amino-terminal 36 amino acids are required for yeast CAP to interact with Cyr1p and activate it in vivo. These differences may simply reflect discrepancies in the activity of the different proteins in the various assay systems.
The activity of Cyr1p complexed with CAP is dependent on the farnesylation state of Ras2p (24)
. However, increased farnesylation of Ras2p does not increase the affinity of Ras2p for adenylyl cyclase (24)
. Mammalian Ras proteins can activate Cyr1p and thereby substitute for yeast Ras2p (25)
. Since GTP-bound Ras2p binds to adenylyl cyclase within a repetitive leucine-rich domain, which shows homology to the leucine-rich repeat family of proteins (26)
, and this affinity is unchanged in the presence of CAP, it was hypothesized that CAP in complex with Cyr1p might constitute a second Ras binding site on Cyr1p. In fact, CAP in complex with the carboxyl terminus of Cyr1p can directly associate with farnesylated Ras, which in turn leads to Ras-dependent Cyr1p activation (27)
.
A screen for randomly mutagenized N-CAP proteins that were defective in RAS signaling but could still interact with Cyr1p has identified three CAP proteins with mutations within the already described 36 amino acid domain (27)
. Two of the CAP mutants were shown to be defective in binding Ras in vitro (N12S/E28G and L13P/E28V) yet maintained the ability to interact with Cyr1p in vitro and in vivo. The N12 and L13 residues of S. cerevisiae CAP are conserved in mammalian CAPs (i.e., N6, L7), once again suggesting this amino-terminal region is important for functional interactions between potential signaling molecules in higher eukaryotes.
CAP is also involved in adenylyl cyclase activation in the fission yeast S. pombe (17)
. Similar to S. cerevisiae, S. pombe cap disruptants exhibit an abnormally large cellular morphology, slower growth, and temperature sensitivity. In contrast, S. pombe cells lacking the amino-terminal of Cap exhibit hyperactivation of mating and sporulation (17)
. The morphological and temperature sensitivity phenotypes can be overcome by expression of S. cerevisiae CAP whereas the sporulation and mating phenotypes cannot, suggesting that adenylyl cyclase and CAP signaling in these two yeasts have functionally diverged. Despite this lack of complementation between yeast CAPs, their amino-terminal domains bind similar proteins, which suggests that, in higher eukaryotes, CAP may interact with related motifs in other signaling molecules yet to be determined. Recently, the amino-terminal domain of Lentinula edodes CAP was shown to interact with a 14–3-3 protein (28)
; the significance of this interaction remains to be determined.
Closer examination of the amino-terminal sequences of CAP reveals the presence of a heptad repeat region (XXXXX; where represents a hydrophobic residue). Heptad repeats are thought to form multiple -helices that wind around each other to form a coiled coil (29)
. Coiled coils are highly versatile motifs involved in oligomerization and protein–protein interactions (reviewed in ref 30
). A Clustal W comparison of all 14 CAP genes characterized reveals extensive homology within this region (Fig. 2
, region A). Within a stretch of 10 amino acids, 7 are present in >80% of all CAP proteins, forming a conserved RLEXAXXRLE motif (where X represents nonconserved amino acid) we term the ‘RLE motif’. The RLE motif is identical to the ‘CAP signature’ motif identified in the ExPASy protein motif database. In yeast, this highly conserved RLE motif interacts with adenylyl cyclase, suggesting its importance. However, the amino-terminal domain of human CAP cannot suppress phenotypes associated with deletion of the same region of CAP in two different yeast species (17
, 31
, 32)
. Therefore, this structurally conserved RLE motif has diverged functionally during evolution but may still be critical for CAP function in all organisms. Perhaps the coiled coil regions in other signaling proteins interact with the CAP RLE motif in higher eukaryotes.
|
The adenylyl cyclase regulatory activity of CAP is also conserved in the yeast Candida albicans (33)
. A CAP1 knockout in C. albicans led to a reduction in cAMP levels and an inhibition of the cells to produce the germ tubes (filamentous growth) required for pathogenicity. Exogenous cAMP addition to cap1-/cap1-cells restored the ability of the knockouts to undergo filamentous growth. It will be interesting to determine whether mutation of the RLE motif in C. albicans Cap1 leads to phenotypes similar to complete gene disruption and retards germ tube formation. Mice infected with cap1-/cap1- cells showed no candidiasis vs. those infected with wild-type C. albicans, which died within 10 days of infection. These results suggest that inhibition of CAP-mediated cAMP signaling pathways may provide an efficacious treatment for candidiasis (33)
.
In conclusion, yeast studies have confirmed the importance of the amino-terminal domain of CAP in adenylyl cyclase activity. Although adenylyl cyclase structure has not been conserved during evolution, it will be interesting to determine whether any aspect of CAP/adenylyl cyclase association has been conserved in higher eukaryotes. If the interaction between CAP and adenylyl cyclases has diverged significantly, then the discovery of other signaling molecules involved in CAP function will be imperative before a clear picture of the function of CAP in higher eukaryotes can be obtained.
Actin binding domain
Initial results suggested that yeast CAP was a bifunctional protein, since disruption of the carboxyl-terminal domain in a wild-type RAS background produced different phenotypes than amino-terminal deletions. Deletion of the carboxyl-terminal domain of CAP resulted in nutritional and morphological defects. Cells showed an abnormally large and round phenotype and an inability to grow on rich medium, and a sensitivity to nitrogen starvation (15
, 18)
. These defects could be rescued by overexpression of the carboxyl-terminal domain of S. cerevisiae CAP (residues 368–526). Cap- cells showed random budding and abnormal actin distribution (34)
. These mutations were complemented by overexpression of the gene encoding profilin (PFY), a monomeric actin and phosphoinositide binding protein (34
35
36)
. The ability of profilin to replace the carboxyl-terminal function of CAP implies a close relationship between these two proteins. Profilin has been shown to interact with proline-rich regions of other proteins, including CAP (37)
. Recently, a large-scale proteomic, two-hybrid screen of 68 yeast proteins (including CAP/Srv2p) against 90% of the predicted open reading frames from S. cerevisiae detected an interaction between CAP and profilin (Pfy1p) (38)
. Further studies are necessary to determine whether this is a direct and physiologically relevant interaction.
Further evidence that CAP can bind actin was demonstrated with the discovery of an actin sequestering protein (ASP-56) from pig platelets that showed a high degree of amino acid identity to mammalian CAP proteins (39)
. ASP-56 (porcine CAP) could bind actin with a 1:1 stoichiometry and could inhibit actin polymerization as measured by falling ball viscometry and fluorescently labeled actin polymerization assays. Similarly, Dictyostelium discoideum CAP has been shown to sequester monomeric actin by inhibiting in vitro actin polymerization in a Ca2+-independent manner with a 1:1 stoichiometry (40)
. This sequestering activity of CAP was restricted to the carboxyl-terminal 210 amino acids; the presence of the amino-terminal 215 amino acids had no effect on actin polymerization (40)
.
S. cerevisiae CAP has been shown to bind G-actin in vitro with a Kd = 0.4 µM, equivalent to the binding coefficient of another actin sequestering protein, thymosin ß-4, to platelet actin (41
, 42)
. Immunoprecipitates of yeast and mammalian CAPs contain actin, suggesting that CAP is bound to actin in vivo (34
, 43)
.
The actin binding activity in the carboxyl terminus of all well-characterized CAPs shows the greatest degree of conservation of any functional domain (Fig. 1)
. Expression of the carboxyl-terminal domain of CAP from one species can complement phenotypes resulting from its deletion in other species (22
, 31
, 32
, 44)
. However, the specific residues involved in actin binding have not been characterized, although a comparison of the carboxyl-terminal domains of all reported CAPs reveals four highly conserved regions (Fig. 2A
, regions D–G). A short deletion of the carboxyl-terminal 27 amino acids eliminates actin binding in S. cerevisiae (43)
and human CAP (45)
. Within this region lies a stretch of 7 amino acids comprising the site E(X)3PEQ (Fig. 2
, region G). The residues E, P, E, and Q are present in all CAP proteins analyzed except the two plant CAPs, which have a substitution of a glutamine for the second glutamate residue. Studies are under way to determine whether these or other carboxyl-terminal residues are critical for actin binding.
Recent experiments with Drosophila CAP have detected a region just downstream from the SH3 binding domain that shows similarity to the verprolin homology domain (LKKAET) (46)
found in a variety of actin binding proteins (e.g., thymosin, fimbrin, and -actinin). Verprolin homology domains are also found in members of the WASp family of proteins, known to bind monomeric actin, and interact with and activate the Arp2/3 complex (47)
. It has recently been reported that actin binding protein (Abp1p), a protein originally isolated from yeast that interacts with F-actin (48)
and CAP through its SH3 domain (49)
, can activate the Arp2/3 complex (50)
. Though intriguing, there is no evidence that CAP participates in Arp2/3-mediated nucleation of actin filaments.
It has been shown that phosphatidylinositol 4,5-biphosphate (PIP2) can promote the availability of monomeric actin for polymerization. Addition of PIP2 at a high molar ratio of CAP to PIP2 (1:40) inhibited sequestration of actin (40)
, suggesting that PIP2 negatively regulates the CAP–actin interaction, causing release of G-actin from CAP and consequently F-actin assembly. The carboxyl-terminal domain alone was unaffected by PIP2 addition, implying that the phospholipid binding site resides within the amino or poly-proline domains (40)
. The negative effect of PIP2 on CAP–actin interaction correlates with the positive effect of PIP2 on activating WASp, which can stimulate actin nucleation by the Arp2/3 complex (51)
. Therefore, the CAP data support a positive role for PIP2 in promoting actin polymerization. However, more studies are needed to determine whether phospholipid regulation of CAP–actin binding is conserved in higher eukaryotes.
Conservation in the carboxyl-terminal domain in all CAPs (Fig. 2
, region D) together with the high degree of conservation in actin structure and function throughout evolution (52
, 53)
suggests that a conserved role in G-actin binding is likely for all CAPs. An important point not yet addressed is whether CAP has differential affinity for specific actin isoforms within the cell and whether the presence of specific isoforms in specific cell types may affect and potentially control CAP function. No information exists on how the interaction between CAP and actin is regulated during activation of signaling cascades.
SH3 binding domain
A third characteristic domain shared by all CAP proteins is a centrally located stretch of poly-proline residues that contains a consensus recognition sequence for SH3 domain containing proteins (PXXPPPXP). In yeast, this domain can be further classified into two regions: P1 and P2. The P1 site is present in all CAP homologues (except D. discoideum) and consists of a stretch of 8–12 amino acids of almost exclusively proline. That profilin binds poly-proline regions and can suppress cap-phenotypes leads to speculation that the P1 region may be a good candidate site for profilin binding. The P2 region contains an SH3 recognition motif in S. cerevisiae, yet this domain is not well conserved in CAPs from higher eukaryotes.
It has been suggested that the SH3 binding site facilitates interaction with a binding partner that targets CAP to sites of actin rearrangement (54
, 55)
. In S. cerevisiae, CAP deleted for the P1 and P2 regions and expressed in a cap strain did not localize to the cortical actin patches as does wild-type CAP. Yeast cortical patches are rich in actin filaments and other cytoskeletal proteins. Even in a cyr1 strain, wild-type CAP localizes to the cortical patches, indicating that correct targeting of CAP is not dependent on its interaction with Cyr1p (54)
. However, the role of the SH3 domain in CAP localization is supplemented by an amino-terminal domain; mutations in the RLE motif (L16P, R19T, and L27F) prevented correct localization of CAP to the cortical patches even when the P2 motif was intact (55)
. The amino-terminal localization domain is distinct from the adenylyl cyclase binding site, since the L27F mutation still interacted with Cyr1p and maintained cAMP signaling. When tagged to GFP, the amino-terminal mutants could only weakly interact with CAP mutants containing the same mutations, implying that the amino-terminal domain is not only necessary for adenylyl cyclase interaction and proper cellular localization, but is important for CAP–CAP interaction (55)
. This could explain why the amino-terminal RLE domain has remained conserved between all CAPs even though homologous adenylyl cyclase enzyme structure has not persisted during evolution.
The poly-proline domain of yeast CAP has been shown to interact with the SH3 domain from the yeast Abp1p (49
, 54)
. Overlay assays using a GST-Abp1p SH3 fusion protein as a probe of various CAP mutants expressed in bacteria demonstrated that mutants possessing only the P2 motif of CAP could retain Abp1p-SH3 binding. Furthermore, abp1 strains showed abnormal CAP localization, although cytoskeletal defects were not detected, suggesting that proper localization of CAP to cortical actin patches is dependent on the presence of Abp1p (49)
. However, the domains of CAP essential for interaction with Abp1p are not required for restoring cell viability to cap strains (49)
. This perplexing result suggests that SH3 domain-mediated targeting of CAP may not be important for its function in yeast or that other proteins may compensate. Another explanation may be that interaction between full-length Abp1p and CAP has not been reported; although yeast genetic evidence and two-hybrid data point toward a connection between Abp1p and CAP, other proteins involved in CAP function and localization cannot be ruled out and probably exist. The lack of conservation within the P2 motif suggests this domain may not be as important in higher eukaryotes.
The SH3 domain of human c-Abl interacted with human CAP in an overlay assay, but in this case the P1 site was necessary for protein–protein interaction (54)
. Since interaction of full-length CAP and c-Abl has not been shown, the significance of this interaction is unclear. However, the important role that c-Abl plays in signaling actin reorganization (56)
implies that an interaction between c-Abl and CAP may have important consequences and be biologically relevant. Further support for the role of Abl in CAP function has recently been reported in Drosophila (57)
. CAP was shown to act antagonistically with Ena, a member of the Ena/VASP family of proteins that catalyze F-actin formation (58)
. Abl inhibited the activity of Ena and thus participates with Ena and CAP to modulate apical actin filament formation in Drosophila follicular epithelium (57)
. Genetic evidence indicates that these two genes have related functions in Drosophila, although whether CAP and Abl interact physically is unknown.
Multimerization domain
Many reports have shown that CAP can form multimeric complexes with itself (45
, 55
, 59)
. Surprisingly, a single dimerization motif has not been defined, although it appears a region in the NH2 terminus adjacent to the adenylyl cyclase binding site in yeast CAP is important for multimerization (55)
(Fig. 1)
. The function of the interaction domain is complex, since two-hybrid screens demonstrate that the amino-terminal domain of human CAP (amino acids 1–228) interacts with itself as well as with the carboxyl-terminal domain (amino acids 253–475). Likewise, the carboxyl terminus interacts with itself and with the NH2 terminus (59)
. This suggests that at least two binding sites exist within CAP that mediate its interaction. One caveat to these two-hybrid results is the presence of endogenous yeast CAP in cells used in the two-hybrid analysis. Since human and yeast CAP can interact with each other (45
, 59)
, yeast CAP could be acting to bridge the interactions between expressed human CAP domains in yeast. The potential interfering properties of endogenous CAP was eliminated by coexpressing a GFP-CAP and an untagged CAP in a cap yeast strain (55)
. Using this in vivo system, an amino-terminal domain was discovered that inhibited CAP multimer formation. Mutations in this amino-terminal domain also prevented proper localization of the protein, suggesting that multimer formation and localization may be linked. Human CAP1 and CAP2, which are 64% amino acid identical, can form heteromeric complexes in vivo that may impart specific functional characteristics yet to be revealed (59)
. It is unclear whether CAP proteins form dimers or higher order structures. A prediction of higher order structures comes from the observation that in fractionation profiles from yeast, CAP eluted between 11.3 and 19.5 S (670 kDa), with higher CAP levels present in the latter fractions (60)
. This suggests that CAP either forms a multimeric structure larger than a dimer or forms stable complexes with other proteins.
|
EVOLUTION OF CAP PROTEINS |
|---|
|
TOPABSTRACTBACKGROUNDSTRUCTURAL AND FUNCTIONAL...EVOLUTION OF CAP PROTEINSADDITIONAL FEATURES OF CAP...CONCLUSIONS AND PERSPECTIVES FOR...REFERENCES |
|---|
. For the purposes of this review, we define a conserved region as a sequence consisting of 7 or more amino acids where more than half of the residues show >80% conservation within the specific stretch of residues. Our definition encompasses motifs that are large enough (>six amino acids) to impart specific function or structure to the molecule and provides an accurate elicitation of protein sequences that are important for CAP structure or function. We are aware that our definition rules out potentially larger motifs (e.g., Pleckstrin homology domains) that arise from secondary structure (e.g., -helices and ß-sheets) and not necessarily from conserved amino acids.
The success of our predictions is demonstrated with region A (Fig. 2A
). Region A encodes a 10 amino acid RLE motif shown to be responsible for adenylyl cyclase binding in S. cerevisiae (24
, 29)
. Deletions within this region also inhibit CAP from forming multimers with itself (55)
. Seven of these amino acids show > 80% conservation between all known CAPs. Homologous region D comprises a region similar to the verprolin homology domain, which has been shown to be involved in actin binding (46
, 61)
. Region G lies within a domain that has been demonstrated to be essential for actin binding (45)
. Therefore, domains responsible for actin binding may be distributed throughout the carboxyl-terminal half of the protein. Region E reveals a highly conserved motif IKGKXNX[I/V] flanked by two conserved cysteine residues. The presence of precisely located cysteines indicates a region of structural integrity that cannot be altered substantially without losing activity. The other three regions (B, C, and F) have not been associated with specific CAP functions. Regions B and C lie within the amino-terminal domain of CAP and are comprised primarily of hydrophilic residues. Region F, adjacent to a known actin binding site, contains a large stretch of homology consisting of several subregions. A search of this sequence using the NetPhos 2.0 phosphorylation prediction program (62)
indicated a potential serine phosphorylation site (score 0.995) within a highly conserved KSSXXN region. Although several potential serine, threonine, and tyrosine phosphorylation sites arise from this type of analysis, only these serine residues (at least one of the two) are present in all CAPs. Studies are under way to determine whether serine phosphorylation occurs and whether this may influence CAP–actin binding. Point mutations within region F that disrupt in vitro actin interaction are currently being studied. We postulate that due to the high degree of conservation within these regions, any perturbation of conserved residues should impair CAP function either through changes in protein–protein interactions or by altering secondary and/or tertiary structure.
A phylogenetic tree containing all CAPs is shown in Fig. 2B
. As expected, CAPs group with their ancestral relatives. All CAPs show higher degrees of amino acid identity within the carboxyl terminus than within the NH2 terminus, suggesting that amino-terminal interactions have not been as conserved. This is supported by the observations that carboxyl-terminal domains from other species functionally complement cells containing carboxyl-terminal deletions of endogenous CAPs whereas deletion of the NH2 terminus is not readily complemented (17
, 31
, 44)
. This implies a functional conservation of actin binding motifs yet a divergence of domains involved in cell signaling. The isolation of signaling targets that interact with higher eukaryotic CAPs will be critical in order to understand CAP function and regulation and to confirm the role of CAP in relaying signals to elements that control cytoskeletal remodeling.
|
ADDITIONAL FEATURES OF CAP PROTEINS |
|---|
|
TOPABSTRACTBACKGROUNDSTRUCTURAL AND FUNCTIONAL...EVOLUTION OF CAP PROTEINSADDITIONAL FEATURES OF CAP...CONCLUSIONS AND PERSPECTIVES FOR...REFERENCES |
|---|
, 54
, 55)
. In higher eukaryotes, CAP is a cytoplasmic protein, but its precise localization is species specific. D. discoideum CAP has been localized near the plasma membranes in resting cells and is remobilized during cell movement (63)
. In GFP-tagged CAP deletion mutants, the amino-terminal domain localized to the plasma membranes whereas carboxyl-terminal and poly-proline containing domains showed a diffuse cytoplasmic staining, indicating that proper localization of CAP is domain dependent (63)
. Dictyostelium cells deficient in CAP showed enlarged cell size and defects in cytokinesis and fluid phase endocytosis.
In mammalian cells, CAP is diffusely distributed throughout the cytoplasm and can concentrate at the cell membrane and lamellipodia of migrating fibroblasts (31
, 45
, 64)
. Human monoclonal antibodies to human CAP1 were recently used to show that human CAP1 colocalized with stress fibers in Swiss 3T3 fibroblasts (64)
. Microinjection of anti-CAP1 antibodies attenuated stress fiber formation in response to serum stimulation and microinjection of purified CAP1 promoted the formation of actin filaments, namely, stress fibers (64)
. Additional experiments are required to confirm the association of stress fibers with human CAP1. Generally, perturbation of CAP levels in mammalian cells appears to influence actin dynamics.
Role of CAP in cell elongation and development
In cotton plants, CAP mRNA has been shown to be highly expressed in young fiber cells vs. other tissues (65)
. Cotton fibers are outgrowths of single epidermal cells from the integument of ovules in the developing fruit. During production of these fibers, individual cells elongate dramatically to >1000-fold longer than their diameter without undergoing cell division (66)
. The cytoskeletal proteins actin, tubulin, spectrin, and the intermediate filament protein vimentin are all present during differentiation, and the dynamic regulation of cytoskeletal architecture is essential for fiber elongation to occur.
Analysis of CAP1 and CAP2 mRNA levels in adult rat tissues reveals a marked difference in expression patterns between the two genes (21)
, which suggests that CAP1 and CAP2 have distinct functional roles and that CAPs are not simply ubiquitous housekeeping genes. The study of CAP transcriptional regulation will undoubtedly shed light on essential functions of CAP in regulating cytoskeletal architecture during development and throughout adult life.
A recent clue about the role CAP proteins play in development has been published from studies of Drosophila (61
, 67)
. These papers have been the subject of a recent minireview (68)
, and some pertinent points will be reemphasized here. Drosophila CAP (named Act Up-acu) was isolated from a screen for mutations that disrupt eye development (67)
. Drosophila cells lacking cap/acu show increased amounts of actin filaments during eye differentiation as well as defects in the formation of the morphogenetic furrow of the eye imaginal disc, which undergoes a dramatic shape change before neuronal differentiation. Drosophila CAP mutants were also isolated that were defective in establishing and maintaining oocyte polarity (61)
. CAP (capulet) was found to be concentrated in the oocyte, where it functions to inhibit actin accumulation. Mutants in protein kinase A (PKA) in Drosophila mirror some of the same cap mutant phenotypes (i.e., loss of nurse cell cortical actin), and actin defects are enhanced in cap pka double germline clones. Therefore, PKA and CAP may be involved in identical pathways that are controlled by cAMP production. It will be interesting to determine whether PKA pathways control CAP activity in vertebrates. The Drosophila studies were performed with mutants that may have been expressing the amino-terminal 97 amino acids of CAP (containing the RLE motif), since computer-generated translations of the Drosophila genomic database suggest that a full-length Drosophila CAP protein is 521 amino acids long and not 424, as published (61
, 67)
(Table 1)
. The 2.4 kb mRNA transcript also supports the presence of a longer transcript more in line with other CAP genes. Whether the presence of the amino-terminal domain affects the interpretation of these results is unclear and should be addressed in future experiments.
Drosophila studies support the conserved role of CAP in eye development and maintaining polarity during early cell differentiation. It is intriguing to speculate that one of the conserved functions of CAP is to control developmental processes that involve cell elongation, migration, movement, and polarity orchestrated by changes in the actin cytoskeleton. On the other hand, CAP plays a role during adult life, since CAP has been shown to be expressed in a wide variety of adult mammalian tissues (21
, 31)
. Sorting out the various CAP functions during pre- and postdevelopment presents one of the more interesting tasks ahead. Since actin polymerization regulation differs significantly in various cell types, it will be important to investigate CAP function in different cellular systems that may or may not require CAP to the same degree.
Role of CAP in vesicle trafficking and endocytosis
The link between the actin cytoskeleton and endocytosis has been well established in lower eukaryotes such as yeast. It is being realized that rearrangements in the actin cytoskeleton may be needed for processes in endocytosis from vesicle formation to vesicle movement (for a review, see ref 69
). Genetic studies in yeast have revealed many genes required for receptor-mediated endocytosis (e.g., END3, SLA2/END4, and RVS167) that, when mutated, cause disruptions in the actin cytoskeleton (for a review, see ref 70
).
Recent studies have elucidated the possible role(s) the actin cytoskeleton plays during endocytosis in mammals. One candidate protein that may link the actin cytoskeleton to endocytosis is mammalian Abp1 (mAbp1). mAbp1 has been shown to bind filamentous actin via its two actin binding domains consisting of the actin-depolymerizing factor homology (ADF-H) domain and a novel actin binding motif (71)
. Immunocytochemical studies show the relevance of the F-actin binding, as mAbp1 colocalizes with the cortical actin cytoskeleton and can relocalize to areas of cytoskeletal rearrangements at the leading edge of migrating cells (71)
.
The first evidence that CAP may be involved in endocytic events was the isolation of a yeast synaptobrevin homologue SNC1 that could partially suppress cap phenotypes (72)
. More recently, yeast CAP/Srv2p has been shown to be synthetically lethal with SLA2 in S. cerevisiae (49)
. Sla2p is essential in yeast and is involved in the cortical cytoskeleton. Sla2p is also involved in endocytosis (73)
and ATPase function at the plasma membrane (74)
. SLA2 is synthetically lethal with ABP1 as well as with SAC6, which encodes the actin filament bundling protein fimbrin (75)
.
More recent studies of mAbp1 have identified it as a putative link between the actin cytoskeleton and endocytosis by its interaction with dynamin (76)
. The interaction between dynamin and mAbp1 was shown to depend on its SH3 domain, and in vivo studies have fortified the relevance of this interaction (76)
. The function of mAbp1 in receptor-mediated endocytosis was demonstrated by a decreased uptake of labeled transferrin upon overexpression of its SH3 domain. Decreased transferrin uptake was abolished on overexpression of dynamin, thus strengthening the significance of the interaction and further supporting a role for mAbp1 as a link between the actin cytoskeleton and endocytosis (76)
.
CAP may link to a dynamin-mAbp1 complex since yeast CAP can interact with Abp1p in yeast. Yeast CAP (SRV2) has been implicated indirectly in endocytic regulation. By screening mutants deficient for endocytosis, a recessive negative form of SRV2 that was unable to internalize pheromone was discovered (73)
. Surprisingly, a mutant bearing a complete deletion of SRV2 was not deficient for endocytosis, suggesting that the mutant form of CAP was causing a disruption of a multiprotein complex (potentially mediated through Abp1p) that inhibited actin regulation and thereby disrupted endocytosis. A characterization of this CAP mutant awaits further study. To complicate matters, End4p/Sla2p was shown to have endocytic functions redundant with CAP and Abp1p (73)
. An END4 mutant missing its coil domain in both a srv2 and abp1 deletion background was unable to internalize labeled pheromone.
The other key player in endocytosis is Rvs167p, a yeast homologue of the mammalian amphiphysin proteins. Amphiphysins are key regulators of endocytosis in mammalian cells (for a review, see ref 77
). Amphiphysins can also interact with dynamin to regulate endocytosis of synaptic vesicles (78)
. Rvs167p can interact with Abp1p and recently was shown to interact with a multitude of yeast proteins involved in actin cytoskeleton and endocytosis in a two-hybrid screen, including Sla2p, CAP, and Act1p (38)
. Therefore, a complex consisting of CAP, Abp1p, Sla2p, and Rvs167p may regulate cytoskeletal turnover during endocytic events. Monomeric G-actin binding proteins like the mammalian CAPs might be involved in areas of high actin rearrangement during endocytosis. It is thought that filamentous actin may organize and hold in place the endocytic machinery; another possibility is that the actin cytoskeleton needs to be broken down in order for endocytosis to occur. Either way, CAPs may play a pivotal role in sequestering monomeric actin, thus aiding in actin rearrangement in these areas. Since mammalian homologues of all these proteins exist, it is reasonable to assume that similar mechanisms probably exist in vertebrate systems.
|
CONCLUSIONS AND PERSPECTIVES FOR FUTURE RESEARCH |
|---|
|
TOPABSTRACTBACKGROUNDSTRUCTURAL AND FUNCTIONAL...EVOLUTION OF CAP PROTEINSADDITIONAL FEATURES OF CAP...CONCLUSIONS AND PERSPECTIVES FOR...REFERENCES |
|---|
). Since the poly-proline domain resides essentially in the middle of the protein, both models allow for the poly-proline SH3 interacting domain to be free to bind target proteins (e.g., ABP1) and render proper localization to the CAP molecule, though other domains may be involved.
|
The regulation of CAP function still presents a mystery. CAP interaction with actin may be controlled through phospholipid binding (i.e., PIP2) in a similar fashion as profilin (Fig. 4
). Phospholipid interactions may regulate the interaction between the amino- and carboxyl-terminal domains. The specific domains within CAP that interact with actin must be defined, and the possibility that actin binding may involve interaction between the amino and carboxyl termini cannot be excluded. We await NMR spectroscopy and X-ray crystallography data on CAP, which will be essential to gain a complete picture of the specific residues important for CAP function.
|
Signaling pathways that control actin reorganization are good candidates for triggering CAP function (Fig. 4)
. In some mammalian cells, the PDGF signaling pathway, which induces membrane ruffling and actin nucleation and cell migration, may be involved (79)
. PDGF can induce relocalization of mABP1 to the periphery of NIH 3T3 cells, which accompanies actin reorganization (71)
. It appears that Ras function between yeast and higher vertebrates has not been conserved since adenylyl cyclase activation by Ras diverged during evolution. However, with the recent observations that yeast Ras may control actin reorganization in response to mild temperature stress and that this effect can be complemented by expression of human Ras, some Ras functions in controlling actin structure may have been conserved (16)
. Although mammalian CAP has not been shown to be involved in RAS signaling directly, any pathway that uses the multiple effectors of Ras, including PI-3 kinase and members of the Rho family of G-proteins that remodel the actin cytoskeleton, are attractive candidates. The interaction of yeast CAP with adenylyl cyclase and its influence on cAMP production imply a role for the PKA pathway in modulating CAP functions. However, the PKA pathway was shown not to be a major player in maintaining cytoskeletal polarity in vegetatively growing yeast (16)
, implying that CAP may have two independent roles: one controlling the generation of specific second messengers and the other involved in actin sequestration. Whether these two processes have direct linkages remains to be seen.
The role of CAP in endocytosis and vesicle transport is intriguing. That CAP can interact directly or indirectly with many key components of the endocytic and actin regulatory network (e.g., Sla1p, Abp1p, Rvs167p, Act1p, and Aip1p) in yeast (38
, 43
, 49
, 54)
suggests an involvement in a large regulatory complex that links elements of the actin polymerization and disassembly network with molecules essential for endocytosis and vesicle movement (Fig. 4)
. Since both processes appear to be intimately linked, further characterization of members within this network will increase our understanding of how actin polymerization communicates with vesicle movement. The ability of mammalian CAPs to interact with homologues of these proteins and participate in vesicle trafficking in higher eukaryotes is now under investigation.
The evolutionary conservation of CAPs suggests an important role in fundamental cellular processes involving actin cytoskeletal reorganization. A better understanding of the role of CAP in these processes awaits further in vivo studies and targeted gene disruptions in higher eukaryotes. Ongoing efforts to discover new CAP interacting proteins will continue, and it will be challenging to demonstrate the in vivo significance of these interactions. Understanding the structural and functional significance of the various CAP domains will be instrumental in our understanding of CAP regulation. Obviously, the ultimate challenge will be to elaborate how this highly conserved protein family functions in diverse biological systems and the role that family members play in linking specific signaling pathways with remodeling of the cellular actin architecture.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTBACKGROUNDSTRUCTURAL AND FUNCTIONAL...EVOLUTION OF CAP PROTEINSADDITIONAL FEATURES OF CAP...CONCLUSIONS AND PERSPECTIVES FOR...REFERENCES |
|---|
