Understanding the function of actin-binding proteins through genetic analysis of Drosophila oogenesis.

Key Words ring canal, actin regulation, actin crosslinker, follicle cell, migration

Abstract

Much of our knowledge of the actin cytoskeleton has been derived from biochemical and cell biological approaches, through which actin-binding proteins have been identified and their in vitro interactions with actin have been characterized. The study of actin-binding proteins (ABPs) in genetic model systems has become increasingly important for validating and extending our understanding of how these proteins function. New ABPs have been identified through genetic screens, and genetic results have informed the interpretation of in vitro experiments. In this review, we describe the molecular and ultrastructural characteristics of the actin cytoskeleton in the Drosophila ovary, and discuss recent genetic analyses of actin-binding proteins that are required for oogenesis.

INTRODUCTION

The production of a Drosophila egg is accomplished through the close association between the developing oocyte and 15 interconnected nurse cells. Transcriptional products from the oocyte nucleus contribute little to the RNA and protein that will sustain early embryonic development. Instead, the nurse cells, and to a lesser extent the follicle cells, synthesize and transport the cytoplasmic components that will support the development of the oocyte and later, the embryo (84).

The mechanism by which cytoplasm moves from the nurse cells to the oocyte has been probed using both cell biological [e.g., (13, 51)] and genetic [e.g., (24, 153)] approaches. Cell biological studies revealed an elaborate actin cytoskeleton in developing egg chambers including extended F-actin bundles and F-actin-rich intercellular bridges [e.g., (109, 145)]. Genetic analysis underscored the functional relevance of the actin cytoskeleton in this process: The majority of mutations characterized at the molecular level that affect nurse cell cytoplasm transport alter ABPs.

Somatic follicle cells surround the germline cysts in each egg chamber, and the follicle cells contribute key signals to determine oocyte and embryo polarity (31). In addition, polarized follicle cells secrete the proteins making up the oocyte vitelline membrane and eggshell late in oogenesis. To carry out these tasks, follicle cells must form and maintain a polarized epithelium, undergo changes in cell shape, and carry out several cell migrations. All of these processes are associated with dynamic changes in the follicle cell actin cytoskeleton.

Thus, the Drosophila ovary provides a system in which the functions of actin-binding proteins (ABPs) can be dissected in a metazoan context. In this review, we focus on recent work on the genetic analysis of actin-binding protein function during Drosophila oogenesis. [For a review of earlier work, see (115).]

Overview of Oogenesis

As in other species, the development of an egg in Drosophila involves complex interactions between germ cells and somatic cells (90, 107). Drosophila ovaries are organized into approximately 15 ovarioles, which are tubular structures that contain progressively maturing egg chambers. Egg chambers initiate development in the anterior of the ovariole, in a region termed the germarium (Figure 1, see color insert). Here, each germline stem cell divides asymmetrically, generating a daughter cell (called the cystoblast) and another stem cell. A cystoblast undergoes four rounds of cell division without completing cytokinesis, generating a cyst of 16 interconnected germ cells. The arrested cleavage furrows in this cluster are subsequently transformed into ring canals, the stable intercellular bridges that facilitate cytoplasm transport throughout oogenesis (112). Interactions between the germ cells and follicle cells in the germarium are complex and not fully understood, but follicle cells are required to maintain ger mline stem cell identity (151), and also regulate germline stem cell proliferation (152). As the 16-cell cyst moves posteriorly in the germarium, interactions with somatic follicle cells continue. Two follicle cell stem cells located at the periphery of the germarium provide a source of follicle cells that encapsulate the germ cells (88). Encapsulation involves complex migratory movements by the follicle cells in the germarium, as the follicle cells send long extensions between adjacent germ cell clusters, resulting in the complete envelopment of the germ cell cluster by a layer of follicle cells (125, 135) (Figure 1B-E). Once this process is complete, egg chambers bud from the germarium and continue to move posteriorly through the ovariole as oogenesis proceeds.

From the cluster of 16 germ cells, one is chosen to become the oocyte. The pattern of synchronized germ cell division in the germarium results in a stereotypical pattern of interconnections, such that only two germ cells have four ring canals, and the oocyte is always selected from one of these two [discussed in (20)]. The oocyte is then localized to the posterior of the egg chamber in a cell-sorting process involving differential cadherin expression in the nurse cells, the oocyte, and the follicle cells (46).

As oogenesis proceeds, the nurse cell nuclei become highly polyploid, whereas the oocyte nucleus remains arrested in meiotic prophase and is transcriptionally quiescent. The increase in nurse cell ploidy facilitates high levels of biosynthetic activity in the nurse cells, and the cytoplasmic products of this activity begin to move into the growing oocyte. Initial transport is slow and selective from stage 1, the point where the egg chamber buds off from the germarium, until stage 10, where the oocyte has grown to a size equal to that of the nurse cell cluster. During stage 10, the egg chambers undergo preparation to rapidly transfer the remaining cytoplasmic contents; this rapid transport phase is initiated by a cue to enter a modified apoptotic program [reviewed in (17)].

The apoptotic cue leads to a number of striking downstream events (17). First, a network of actin bundles forms in the nurse cells, creating a halo of F-actin around each of the large polyploid nurse cell nuclei. This is followed by the disassembly of the nurse cell nuclear envelopes, which allows the nurse cell nuclear contents to mix with cytoplasm. Finally, the nurse cells actively contract, expelling their remaining cytoplasmic contents into the oocyte. The cytoplasmic bundles are required to restrain the nurse cell nuclei during rapid transport, as mutations that compromise bundle formation still contract, but fail to transfer cytoplasm due to the movement of the large, polyploid nuclei into the ring canals, blocking further transport [e.g., (24)]. Based on several observations, the contractile force appears to be generated by cytoplasmic myosin II. Germline clones of sqh, the gene encoding the myosin II regulatory light chain, fail to transport due to a failure in contraction, and Myosin II is localized to the nurse cell cortices (34, 149).

Follicle cells undergo a precisely regulated series of migrations during oogenesis [reviewed in (31)]. Beginning at stage 9, most follicle cells migrate as an epithelial sheet posteriorly toward the oocyte. About 50 follicle cells remain covering the nurse cell cluster and become stretched into a squamous epithelium. During the posterior migration of the follicular epithelium, a special population of follicle cells, the border cells, delaminate from the anterior of the egg chamber and move through the nurse cell cluster to the anterior margin of the oocyte. These cells are required to form a functional micropyle, a structure that facilitates sperm entry [reviewed in (126)].

The Actin Cytoskeleton in the Germline

The organization of the actin cytoskeleton in the Drosophila ovary has been extensively characterized in terms of ultrastructural organization and molecular composition. We briefly review what is known about the ovarian actin cytoskeleton, and compare the actin-rich structures in the ovary with common organizational features of the mammalian actin cytoskeleton. In ovarian germ cells, two striking populations of actin filaments exist: ring canal actin, and stage-specific cytoplasmic actin bundles.

The cytoplasmic actin bundles that form just prior to the rapid phase of cytoplasm transport bear a number of striking similarities to parallel actin bundles in other systems, including brush border microvilli and hair cell stereocilia [discussed in (5, 29)]. The cytoplasmic bundles are composed of unipolar, hexagonally packed actin filaments with their fast-growing barbed ends facing the membrane (Figure 2A and C, F and G, see color insert) (49). At least two filament crosslinking proteins are required to organize these bundles. Quail, a relative of brush border villin (85, 91), appears to loosely bundle filaments, while fascin, the product of the singed gene (19), organizes these filaments into maximally crosslinked, hexagonally packed filament bundles (19). Actin bundles exhibit a striated appearance when visualized with fluorescent phalloidin (49, 109) (see Figure 2C). The striated appearance likely reflects their organization, as the bundles, which extend up to 40 [micro]m in length, are composed of a se ries of overlapping 2-3 [micro]m modules. These modules appear to be attached to each other through lateral associations, and the gaps in fluorescence likely reflect regions of little overlap of adjacent bundles (49) (Figure 2G).

Fluorescent-phalloidin staining of ring canals reveals a tight band of F-actin (Figure 2B). Images of ring canals in cross section obtained by thin section EM show an electron-dense outer rim adjacent to the ring canal membrane and an electron opaque inner rim that corresponds to the F-actin visible by phalloidin staining [e.g., (140)] (Figure 2D, E). By EM, the F-actin at the ring canal appears to be organized into loosely packed parallel bundles throughout most of oogenesis (140) (Figure 2E). A number of proteins have been found by immunofluorescence to localize to ring canals, and a distinction can be made between ring canal proteins that colocalize with F-actin in the inner rim and proteins that are more closely associated with the membrane in the ring canal outer rim.

It is difficult to make a perfect analogy between the ring canal and other F-actin structures. However, an intriguing comparison can also be made between ring canals and the leading edge in a migrating cell. Perhaps the most telling similarity is the high rate of actin filament turnover in both ring canals (69) and leading edge lamellipodia [e.g., (136)]. This high rate of F-actin turnover is in marked contrast to the F-actin associated with stress fibers [(158) and references therein]. In addition, ring canals share several molecular components with lamellipodia, including filamin and the Arp2/3 complex (discussed below).

The Actin Cytoskeleton in Follicle Cells

Ovarian somatic follicle cells adopt markedly different morphologies and functions during oogenesis, providing an excellent model for a number of actin-dependent processes. The follicle cells that surround the young egg chambers and cover the oocyte in later egg chambers provide a beautiful example of a polarized epithelium (Figure 3A, see color insert). This includes what appears to be an actin-rich adhesion belt near the apical surface and clear apico-basal polarity [reviewed in (100)].

The follicular epithelial cells also have an array of actin bundles associated with their basal plasma membranes, which is in contact with an overlying basement membrane (52,53) (Figure 3D, E). These bundles may be analogous to stress fibers, and they are required for egg chamber shape. Late in oogenesis, egg chambers change in shape from round to oblong. This shape change is preceded by a reorganization of the basal actin bundles: Basal bundles in younger egg chambers appear randomly oriented from cell to cell, but by stage seven become aligned such that bundles in all cells lie perpendicular to the anterior-posterior axis of the egg chamber (40). In mutants where the bundles are disorganized, such as kugelei, the resulting egg chambers fail to elongate properly and the resulting eggs are round and sterile (53). Given this phenotype, it seems plausible that the bundles are contractile, and that they are thus bona fide stress fibers. However, it has not been determined whether myosin is a component of these b undles, nor has the polarity of the filaments in the bundles been determined.

Near the end of oogenesis, the border cells begin their migration as a group of 8 to 10 follicle cells at the anterior of the egg chamber that delaminate from the follicular epithelium (Figure 3G). In the course of their migration, the border cells exhibit all of the stereotypical features associated with migrating cells: lamella and lamellipodium extension, filopodial extension (see Figure 3G), and cell translocation. Thus, border cell migration presents the opportunity to study the actin cytoskeleton in cells migrating in situ in a genetically tractable system.

PROTEINS AFFECTING THE POLYMERIZATION STATE OF ACTIN

Understanding how cells rapidly reorganize their actin cytoskeleton requires the identification and characterization of proteins that influence actin polymerization dynamics. In vitro, the initiation of actin polymerization (nucleation) is an inherently slow process; however, once polymerization is initiated, it proceeds rapidly until a steady state between G-actin and F-actin pools is achieved. Proteins that initiate, terminate, or otherwise affect the rate of actin polymerization are likely to be important regulators of actin polymerization in vivo. In recent years, the study of additional proteins involved in actin polymerization dynamics in Drosophila has added to our understanding of the actin cytoskeleton.

The Arp2/3 Complex and its Regulators, Scar and Wasp

The Arp2/3 complex is a complex of seven proteins first identified as a profilin binding complex in Acanthamoeba (78). It consists of the actin-related proteins Arp2 and Arp3, a WD-repeat protein Arpcl, and four other proteins, Arpc2-5 (79). The Arp2/3 complex is the only actin nucleation factor thus far characterized that creates new actin filaments that elongate from their barbed ends (101). As most actin polymerization within a cell involves growth at the faster-growing barbed end, the Arp2/3 complex is positioned to be an important regulator of actin polymerization. The Arp2/3 complex requires two cofactors for efficient nucleation: an activating protein of the Wasp/Scar family and preexisting actin filaments. Current data indicate that the Arp2/3 complex binds to an activating protein and to the side of an actin filament, and this activated complex mimics the structure of a barbed end of an actin filament (117, 144). The binding to the side of a filament is thought to be obligatory for Arp2/3 nucleation (3), and this leads to the polymerization of new filaments at a characteristic 70[degrees] angle from the mother filament (101). In agreement with this, the Arp2/3 complex has been localized to 70[degrees] filament branches in lamellipodia of cultured cells (37, 132).

Scar and Wasp activation are downstream of the Rho-family of small GTPases. In the case of Scar, there is evidence that the Rho family GTPase Rac causes changes in the actin cytoskeleton through Scar (80, 96), and that Scar in turn activates the Arp2/3 complex via an intermediary protein, IRSp53 (97). However, the best-documented signaling relationship involves actin polymerization downstream of the small GTPase Cdc42. Cdc42 binds and activates Wasp, which in turn activates the Arp2/3 complex, leading to new actin polymerization [for review, see (59)]. Scar and Wasp are homologous in their C-terminal regions that mediate interaction with the Arp2/3 complex, but have divergent N-terminal domains that interact with upstream signaling molecules (Figure 4, see color insert).

Recent isolation of mutations in the Drosophila homologs of Scar, Wasp, and Arp2/3 complex components has…