Effect of Cdc42 domains on filopodia sensing, cell orientation, and haptotaxis

Highlights

• Cells orient more filopodia toward the adhesive end of a haptotactic gradient.

• Filopodia are the only protrusions involved in comparing adhesiveness across a cell.

• Sensitivity to the gradient is proportional to the prevalence of filopodia.

• Cdc42 domains binding to Par6, IQGAP, and ACK are critical for sensing adhesiveness.

• Other protrusions respond to the gradient only if filopodia are high in prevalence.

Abstract

Filopodia are sensors which, along with microtubules, regulate the persistence of locomotion. To determine whether protrusions were involved in sensing adhesion, epithelial cells were cultured on platinum and tantalum gradients. Protrusions were defined by an unbiased statistical method of classification as factors 4 (filopodia), 5 (mass distribution), and 7 (nascent neurites). When the prevalence of protrusions was measured in zones of high (H), middle (M), and low (L) adhesiveness, the main differences were in factor 4. Its values were highest at H and declined at M and L regardless of the gradient composition. The significance of the differences was enhanced when T (top/adhesive end) and B (bottom/nonadhesive end) sides of cells were analyzed separately. Since information about sidedness increased the statistical power of the test, this result suggested that cells pointed more filopodia toward the adhesive end. Trends occurred in factors 5 and 7 only when conditions allowed for a marked trend in factor 4. The data showed that gradient sensing is proportional to the prevalence of filopodia, and filopodia are the only protrusions engaged in comparing adhesiveness across a cell. The probability (P) of the significance of a trend was then used to determine how cells sense the gradient. Binding peptides (BPs) were introduced representing sequences critical for Cdc42 docking on a specific partner. BPs for IQGAP (IQ(calmodulin-binding domain)-containing GTPase-activating protein) and ACK (Cdc42-associated kinase) reduced factor 4 values and prevented cell orientation on the gradient. Micrographs showed attenuated or stubby filopodia. These effectors may be implicated in gradient sensing. Another IQGAP BP increased filopodia prevalence and enhanced orientation on the gradient (P < 0.00015). A Wiskott–Aldrich syndrome protein (WASP) BP had no effect. When sensing and orientation were abolished, they both failed at the level of filopodia, indicating that filopodia are both sensors and implementers of signals transduced by adhesion.

Introduction

Filopodia play a role in motility. They are considered ‘antennae’ of the cell, which extend and retract to explore features of the environment [1]. In some instances, for example during dorsal closure in embryonic development, where filopodia can be viewed in situ, they are found at the leading edge of the cell. In the axon growth cone, the direction of persistent locomotion is set by filopodia [2]. This laboratory found that the prevalence of filopodia declined in cancer cells [3], [4]. Thus, it is possible that cells depend on filopodia to become oriented in the environment, and cancer cells become disoriented by the loss of filopodia. The loss of cell polarity contributes to the tissue disorganization which is considered a hallmark of cancer [5]. Conversely, mechanisms contributing to the establishment and maintenance of polarity are required to organize cells into tissues. It is possible, therefore, that filopodia-mediated sensing is important in growth control.

Certain mechanisms of polarization are primitive and are exhibited by single-celled organisms. In the budding yeast, Saccharomyces cerevisiae, the bud is formed at a pole established by a landmark complex of proteins. The Ras homolog, Rsr1p/Bud1p, is recruited to the cell cortex by the landmark complex. Bud1p in the GDP-bound state is sequestered by Bem1p. Upon activation, Bud1p binds to two cell division cycle (Cdc) proteins, Cdc24p and GDP-bound Cdc42p [6], [7]. Cdc42 then establishes an axis of polarized growth by triggering an asymmetric organization of the actin cytoskeleton and secretory apparatus. Bem1p is one of the classes of scaffold proteins that have no other function but bringing together components of a signaling module. Several components of the yeast bud site have orthologues in the metazoan cell, including Cdc42, Bud1p, and Cla4p. Thus, there is reason to think that the core of the yeast bud organizer is conserved in the focal contact of mammalian cells (Fig. 1). Whereas Cdc42p is concentrated at one site in the cortex of the yeast cell, its pattern of localization on mammalian cell membranes is diffuse. These cells interact with the microenvironment through a variety of distinct protrusions, for example, ruffles, lamellipodia, filopodia, and neurites. Most of these protrusions are anchored to substrata by focal contacts or focal adhesions. Although these structures are implicated in cell polarity and/or orientation, it has been difficult to assign specific functions to each of their protein components. This problem is especially acute for scaffold proteins, because they can be upstream of a whole network of other protein components. For Cdc42, which is highly conserved through evolution, it is possible to harness knowledge from molecular dynamics to study its function, because many Cdc42-protein interfaces have been characterized by x-ray crystallography or mutational studies of binding.

Previously, the laboratory developed methods for the unbiased classification of cell features, based on the use of size-invariant measures of shape [8]. Using a standard statistical procedure, we could extract latent factors that correspond to cell features. Latent factors are unobservable variables that correspond to the co-variance of two or more measurable variables [9]. It is one of the several standard statistical methods used to reduce the redundancy of image content in large image databases. Features specific to the edge represented protrusions [3], [4]. It was clear that factor 5 quantified mass displacement from the central portion of the cell, a geometry that is well-understood [10]. Earlier work showed that factor 4 represents filopodia [11] and factor 7 is a nascent neurite [12]. Since features could be identified quantitatively and qualitatively, the method made it possible to investigate how cells use their protrusions to sense and interact with the environment. Moreover, differences in features across the width of the cell could be measured. We have investigated the role of Cdc42 and filopodia sensing in haptotaxis. By doing so, we also elucidated mechanisms of gradient sensing at a single cell level. Since only single cells were analyzed in the experiments, the results reflect the way in which cells compare adhesiveness on opposite sides of the cytoplasm. This approach allows us to study the function of scaffold proteins, for example, IQGAP, which are components of focal complexes. Moreover, it is a method whereby the function of scaffolds can be approached, which is otherwise difficult because interference with the platform function necessarily perturbs all downstream pathways. IQGAP has a cross-linking role for actin and microtubule elements (see Section 4.2) and is specifically implicated in receptor activation and in processing of vascular endothelial growth factor, CD44 (hyaluronic acid receptor), epidermal growth factor (EGF), and fibroblast growth factor [13].

Haptotactic substrates were originally created by depositing a gradient of metal on a nonadhesive material. Mammalian cells move in a directional manner up the gradient toward the more adhesive end [14]. The metal substrate is adhesive because of the adsorbed proteins, and indeed, similar gradients have been formed by depositing a purified protein such as collagen on a nonadhesive substrate [15]. In the conventional haptotactic experiment, the nature of the extracellular matrix and the distribution of its components affect the speed and direction of cell migration [16]. The term was also used by workers to describe the more complex process of cell migration through porous membranes, however. Because filopodia mediate the persistence of locomotion in mammalian cells, we postulated that they would point to the adhesive end of the gradient. The specific role of Cdc42 in filopodia is controversial, however. It was thought to bind to Wiskott–Aldrich syndrome protein (WASP) and help activate Arp2/3-mediated filament formation in filopodia [17], [18], [19], but later work suggested that this pathway of filopodia formation was uncommon (see [2] for review). Nevertheless, as mentioned above, there is reason to think that Cdc42 was instrumental in maintaining polarity in primitive biological organisms. If Cdc42 were implicated in filopodia dynamics or persistence, and thereby mediated directional locomotion, parts of the molecule would be indispensable for filopodial sensing and/or signal transduction. This postulate was tested by introducing into the cells BPs that mimicked short sequences of Cdc42. Such sequences were designed to represent domains where docking occurs on binding partners in situ. By mass action, the newly introduced BP can interfere with Cdc42's binding to its partners, and an effect on filopodia prevalence can be measured, if indeed Cdc42 plays a role in filopodia dynamics.