Dictyostelium amoebae and neutrophils can swim

Abstract

Animal cells migrating over a substratum crawl in amoeboid fashion; how the force against the substratum is achieved remains uncertain. We find that amoebae and neutrophils, cells traditionally used to study cell migration on a solid surface, move toward a chemotactic source while suspended in solution. They can swim and do so with speeds similar to those on a solid substrate. Based on the surprisingly rapidly changing shape of amoebae as they swim and earlier theoretical schemes for how suspended microorganisms can migrate (Purcell EM (1977) Life at low Reynolds number. Am J Phys 45:3–11), we suggest the general features these cells use to gain traction with the medium. This motion requires either the movement of the cell’s surface from the cell’s front toward its rear or protrusions that move down the length of the elongated cell. Our results indicate that a solid substratum is not a prerequisite for these cells to produce a forward thrust during movement and suggest that crawling and swimming are similar processes, a comparison we think is helpful in understanding how cells migrate.

The leading front of an animal cell migrating over a substratum is differentiated from the rest of the cell by several properties believed to be linked to how the cells migrate (1, 2). It is the main site of actin filament formation (3) whose polymerizing force may impinge on the membrane at the front of the cell to force it forward and so advance the cell’s front (4–7). It is also the site of exocytosis of recycling membrane from the cell’s internal pools (8) whose area could provide the surface membrane required for the cell to extend itself forward (9). Whatever the actual process, as a cell moves up a chemotactic gradient the information which the cell’s surface receptors collect to steer the cell toward the source must determine which part of the cell’s surface constitutes the front and must activate a motor there to extend the cell toward the source.

Cell migration and chemotaxis generally are studied as the cells crawl over various solid substrates. The possibility that amoebae might swim arises from the observation that a mutant of Dictyostelium discoideum, sadA, which attaches poorly to a substrate, appears nevertheless to migrate normally and does so with an enhanced speed. The product of this gene is a plasma membrane protein which, when expressed in sadA, restores normal attachment and behavior (10). However, the extent to which the sadA strain is able to form transient attachments to a substratum is unclear. We therefore sought to discover whether Dictyostelium amoebae require a substrate upon which to move or whether they can swim.

Results

To find whether cells might move in the absence of a substrate, we tried to find whether amoebae could migrate at a liquid–liquid interface or an air–water interface. With Christien Merrifield (of this laboratory), we found that these cells indeed can move at both interfaces, as had many others (e.g., ref. 11). However, the difficulty in interpreting such observations is that we do not know, and have no means of knowing, what lies at any interface or whether there may be some physical peculiarity there with which the cells could be gaining traction.

We therefore sought conditions in which there was no interface, in which gravity would not play a dominant role, and in which cells would survive and not be subject to an extra osmotic load. We found that amoebae suspended in 10–12% Ficoll settle out very slowly. Initial observations in a chamber (in a 96-well plate) showed that convection obscured any movement of the amoebae that might have occurred, despite the dampening effect of the increased viscosity of this solution (about 8 cPoise (12), compared with 1 cPoise for water). To overcome this difficulty, we designed a smaller observation chamber 5 mm deep and 5 mm wide (Fig. 1 and Materials and Methods) into which several layers of Ficoll were added above a base of 20% Ficoll, creating a density gradient for enhanced stability. Because of the substantial diffusion of Ficoll, such gradients last only for several hours, after which the gradient is lost, and the cells settle out. In addition, we used developing amoebae that chemotax toward a source of cAMP so that we could look for directed, as opposed to random, movement. A point source of cAMP was provided by a needle introduced through the open top of the chamber; this needle in turn was held in place by a support rigidly attached to the chamber. To reduce surface evaporation, the chamber was covered with a layer of mineral oil. Fig. 2A shows a time series of images of the focal plane of amoebae, suspended in this medium, and the tip of a needle through which cAMP is leaching. The cells migrate toward the chemoattractant source, which itself is about 1 mm from the bottom of the chamber. Many of these cells collect and aggregate around the needle tip. After the final frame of the experiment shown in Fig. 2A, we took a z-stack of images (Fig. 2B) to examine the distribution of amoebae around the needle tip. This z-stack shows that a three-dimensional cloud of cells has collected around and aggregated on the tip. The movements seen in Fig. 2 A and B show that the cells have translocated toward and concentrated around the needle tip from all directions.

Fig. 1.

Chemotaxis chamber.

Fig. 2.

Chemotaxis of amoebae toward a cAMP source. (A) Developing Ax2 cells were suspended in KK2 buffer containing 10% Ficoll and placed in a Ficoll gradient in the chamber. A needle loaded with 0.1 mM cAMP was placed in the chamber so that its tip was about 1.5 mm from the base of the chamber. Cell movement was followed in an inverted microscope using a 5× objective taking images in the plane of the needle tip every 20 s (Movie S1). The cells can be seen to move toward and to congregate around the tip. Times are indicated in minutes. (Scale bar, 0.1 mm.) (B) z-Stack taken at the end of the experiment shown in Fig. 2A. A series of images was taken from −300 μm to +300 μm in 15-μm steps (Movie S2); every fifth image is shown. The position below or above the needle tip is indicated. The cells crowd around the needle tip, showing that their motion is in three dimensions. (Scale bar, 0.2 mm.)

In a control experiment, no such migration occurred if the needle did not contain any chemoattractant. This experiment shows that the movements described are chemotactic and not driven, for example, by a flow of medium toward the needle tip. In a separate recording at higher magnification at which individual cells are seen more easily, we found that the average linear speed of the cells toward the tip (measured as the crow flies) is about 4.2 μm/min (a total of 23 cells measured over an average of about 16 min had a range of 2–8.4 μm/min during this time). This speed is similar to that of cells attached to a glass surface in the same medium (about 3.8 μm/min). This speed is about a third of that typically seen for developing cells in a normal environment, a decrease that we attribute to the greater force needed to move through the higher viscosity of the medium. These results indicate that amoebae can chemotax is this medium.

To gain a greater understanding of how these cells move, we used a cell line expressing high levels of the plasma membrane marker cAR1-GFP (13) to view the surface outline of the cell. We used a spinning disk confocal microscope, collecting images at a rate of a stack every 10 s using a 60× objective focused on the volume next to the needle tip and then projecting the stack into a single plane. Fig. 3 presents a panel of such frames at 10-s intervals for a typical cell. The cell is extraordinarily active, changing its shape quickly and producing transient spikes and membranous veils as it extends itself forward. It frequently bifurcates or produces side lobes.

Fig. 3.

The shape of amoebae as they swim. An amoeba whose surface was labeled with cAR1-GFP chemotaxes toward the cAMP source (just above the top of the images). z-Stacks of images (from −9 μm to +9 μm in 3-μm steps) were collected every10 s, and the images were collapsed into a single plane. Shown is a series of consecutive images with the times indicated in seconds (taken from Movie S3). Because each image was collected over a period of about 3 s, images shown are not true instantaneous volume images. The cell changes shape surprisingly quickly; arrowheads indicate features that apparently can be traced from one image to the next. (Scale bar, 10 μm.)

The speed with which these features appear or are modified means that features seen in one frame cannot be identified unambiguously in succeeding frames. However, as indicated by the arrows in a few frames that tentatively identify what we believe to be the same feature in successive frames, spikes and lobes move rearwards, as they do on cells migrating on a substrate.

Neutrophils Chemotaxing Toward a formyl-Methionyl-Leucyl-Phenylalanine Source.

To see whether this ability to migrate in suspension might be a wider property of crawling cells we studied human neutrophils in a similar way. Because neutrophils have a somewhat greater density than amoebae, we increased the density of the Ficoll layers with an additional ~9% Histodenz in the medium. Chemotaxis toward a point source of the tripeptide formyl-methionyl-leucyl-phenylalanine (fMLP) was followed at about 24 °C (unhelpful convection currents were greater at 37 °C). Fig. 4 shows neutrophils migrating toward the needle tip. By the end of the experiment, a large blob of cells has become attached to it.

Fig. 4.

Chemotaxis of neutrophils toward an FMLP source. Fresh neutrophils were suspended in RPMI medium containing 5% FCS, 10% Ficoll, and about 9% Histodenz (a small molecule that increases the medium density). This mixture was incorporated in a density gradient in the chamber, and the motion of the cells was observed as in Fig. 2A, using a 10× objective. Polymorphonuclear leukocytes migrate to the needle tip where they attach, forming a large blob. In a control experiment without FMLP in the needle, chemotaxis did not occur. A z-stack (taken as in Fig. 2B) indicated that the needle tip is far from the base of the chamber. Images were collected every 30 s for a total of 200 min (see Movie S4). Times are indicated in minutes. (Scale bar, 20 μm.)

We tried to estimate the speed with which neutrophils move under these conditions. Many cells do not chemotax, so our data have been selected for those cells that move toward the source. These data yield a speed (measured in a direct line from start to end) of about 2.4 μm/min (20 cells, measured over an average of about 40 min, range 1.1–4.1 μm/min). Again, this speed is similar to the speed of these cells migrating on a glass coverslip under similar conditions (~1.6 μm/min). The migration of neutrophils toward the needle tip is not as impressive as that of amoebae. Neutrophils are slower, presumably because they normally move at 37 °C, and the distance at which they react to the chemoattractant is shorter, perhaps because amoebae not only sense the needle source of chemoattractant but then signal to one another, thereby amplifying the signal.

Discussion

Our observations show that amoebae and neutrophils suspended in a Ficoll solution are able to chemotax. The cells clearly migrate in three dimensions toward the chemoattractant source (Fig. 2B), which itself is far removed from the substratum on the base of the chamber. How do they achieve this movement? It might be imagined that they settle at an interface between the 10% and 20% Ficoll layers and somehow use the interface to migrate. However, such an interface has no rigidity on which to gain traction. Indeed, by the time these observations were made (at least 20 min after the layers had been assembled), any interface would have formed a gradient by diffusion. Another possibility is that, at the high Ficoll concentrations used here, these polymers associate to form a matrix or gel within which the cells could gain traction. Earlier studies of the physical chemistry of Ficoll solutions show that, even at high concentrations, individual molecules have a diffusion coefficient typical of a molecule of that size (~10⁻⁷cm²/s) (14). Thus the Ficoll solutions we use here have no extensive lattice. Furthermore, at low concentration, Ficoll molecules behave like hard spheres (15), as would be expected from a globular (as opposed to a linear) polymer. In addition, the viscosity of Ficoll solutions increases monotonically with concentration (determined with Ficoll 70) (16), further indicating that no extensive meshwork or lattice is formed. The possibility that amoebae produce a slime matrix that enables them to move [as do some bacteria (17)] is improbable: A main component of slime produced by amoebae is cellulose, which, like all known slime proteins, is not synthesized until much later in development (18). We conclude that amoebae and neutrophils can swim. If leukocytes also can swim, this ability may explain how these cells are able to move through a three-dimensional matrix without any apparent need for integrin molecules with which to bind to the matrix (19).

What does this information indicate about how amoebae migrate over a normal substratum? It could be argued simply that the basic mechanisms by which these cells crawl on a surface and swim are different—after all, the manner in which people walk or swim is inherently different. However, although this possibility cannot be ruled out at present, it seems unlikely. In both cases, cells move similarly: They project forward a leading lamella or lobe and sometimes microspikes. When the front bifurcates, one lobe recedes and, as it does so, moves backward on the cell together with other projections on the cell’s surface. In both circumstances the cells move with similar speeds. Furthermore, in both situations—whether migrating on a substrate or swimming in suspension—they respond to a chemotactic agent: The machinery that interprets this gradient and orients the movement of the cell must act on force generators that drive both migration and swimming. The similarities in speed and phenotypic and chemotactic behaviors suggest that the same prime mover effects both crawling and swimming.

When animal cells migrate over a substrate, they often show an enrichment of F-actin at their fronts (20). The polymerization of G- to F-actin has been suggested to effect the actual protrusion of the leading edge as the cell advances (4–7). If these filaments are transiently anchored in some way to the substrate, this scheme is logically coherent. However, such polymerization at the front could not help a swimming cell move forward: The front of the cell would advance, but its rear would move backward at the same time. In other words, the centroid of the cell would remain stationary, because no traction with the environment could be achieved.

In 1977, Purcell summarized the different mechanisms by which microorganisms might be able to swim (21). In what is now known as “Purcell’s scallop theorem,” he points out that, because of the low prevailing Reynolds number, viscous forces predominate such that momentum is irrelevant and movement requires a nonreciprocal motion. Thus, a microscallop, by rapidly closing its shells and slowly opening them—a process that is reciprocal in time—would not move. The principal difficulty the cell faces is that it must push the medium backward to change the position of its own center of gravity and do so in a manner that satisfies Purcell’s theorem. Purcell discussed at least three different ways in which cells can swim (Fig. 5), which we summarize as

• (i) They use a rotating screw to force the medium backward. This method is how bacteria (22, 23) or ships move.

• (ii) They engage in a series of asymmetrical shape changes. This method would include flexible oars, such as cilia, or breast stroke to push the medium backward.

• (iii) In a toriodal cell with a rotating surface, the medium would be dragged backward by the moving outer cell surface, providing the thrust to advance the cell.

Fig. 5.

Schemes for how cells might swim. (A) Rotating screw. (B) Flexible oar. This scheme covers a variety of complex movements involving any series of nonreciprocal shape changes. (C) Cross-section through a toroidal cell. (D) Circulating membrane through the cell’s interior. (A–C are taken from ref. 21; D is adapted from ref. 28.)

Because neither amoebae nor neutrophils have flagella, the first scheme cannot apply. These cells change their shapes rapidly as they move, but the projections they produce appear to move backward rather than flex about a fixed point typified in breast stroke or ciliary motion (the second scheme). However, we think a modified version of scheme iii is closer to what is observed. There are two aspects to this mode: The central hole (Fig. 5C) could be vanishingly small and membrane internalized into the cell on one side and transferred through the cell to the other side (Fig. 5D), generating a flow equivalent to that indicated in Fig. 5C and so moving the cell. Such a surface flow could be generated by a polarized endocytic cycle (9, 24), although there is evidence that such a flow does not exist in amoebae (25) but does in neutrophils (26). Alternatively, cellular projections—whether spikes or lobes—could form at the front of the cell, move rearwards along the cell, and be resorbed near its tail and so provide the thrust to propel the cell forward in a process that is a hybrid of the second and third schemes. We believe that these projections and/or a rearward surface flow are responsible for providing the force against the medium that is used by the cells to swim.

These experiments suggest that the manner in which cells attach to a substratum upon which they move is in itself not essential for movement and that these attachments actually may serve as a brake on the cells’ speed, as suggested for sadA by Fey et al. (10). They show that animal cells can swim and suggest a different way of viewing cell migration: We think it helpful to think of crawling on a substrate and swimming as two aspects of the same basic process.

Materials and Methods

Dictyostelium discoideum amoebae (Ax2 or cAR1-GFP) (13) were grown at 22 °C in axenic medium, and development was initiated by pulsing with cAMP for 3.75 h (25), after which they were resuspended in fresh KK2 buffer (20 mM potassium phosphate, 2 mM magnesium chloride, 0.1 mM calcium chloride, pH 6.1). Neutrophils, prepared from the blood of volunteers using Percoll gradients and kindly provided by members of the Department of Medicine, University of Cambridge, were held in PBS containing 10% autologous serum at room temperature at 2 × 10⁷ cells/mL until used.

Chemotaxis Chamber.

A small, circular glass chamber was constructed to examine the chemotactic behavior of cells in suspension. This circular chamber (5 mm wide × 5 mm deep) was cut into a thick 2.5 × 2.5-cm slide. One side of the top of the chamber had a 3-mm 45° beveled port to enable a glass needle to reach the chamber bottom at the middle of the chamber. Finally, a glass coverslip (13 mm, #0) was sealed to the chamber’s base with Araldite (Huntsman Advanced Materials Ltd). For microscopy, the chamber was held in a frame on the microscope stage: The square slide with the chamber was placed into a circular hole in this frame and wedged there securely with sections of yellow tips (200 μL; SARSTEDT). Above the top level of the chamber a second glass slide (0.5 × 2.5 × 2.5 cm) was fastened securely to the frame with double-sided adhesive. On top of this stage was a plastic ramp (an isosceles triangle in cross-section with a short side of 1.5 cm and depth of 1 cm) with an angular groove set into its hypotenuse in which a needle could be slid up and down. This ramp was fixed to the glass stage with Araldite. The position of the needle with respect to the chamber could be varied by undoing the joint with the double-sided Sellotape. Needles were held in place with a clip.

For light microscopy, millipored solutions of Ficoll (Ficoll 400, cell-culture tested, F-8015; Sigma) in KK2 buffer were layered sequentially in the chamber. A base, usually 20 μL 20% Ficoll; a cell layer of 20–30 μL of 10–12% Ficoll; 20 μL 5–8% Ficoll; 20 μL 0% Ficoll; and finally about 40 μL mineral oil (M-3516; Sigma) to prevent evaporation during the experiment were layered sequentially in the chamber. The second layer was approximately isodense with the test cells, about 10% Ficoll for amoebae. This density was estimated empirically by suspending a few cells in increasing concentrations of Ficoll in KK2 buffer in a 96-well plate and centrifuging the suspension for 3 min at 300 × g to find concentrations at which cells both settled and floated. For neutrophils, the same procedure for finding an isodense medium was followed except that all tests were done in 10% Ficoll in RPMI medium 1640 (GIBCO) buffered with Hepes to pH 7.0 but with increasing concentrations of Histodenz (D-2158; Sigma), maintaining an osmotic balance using an isotonic 26.7% solution of Histodenz (27). All buffers with neutrophils were based on RPMI medium and isotonic Histodenz in a 0.64:0.36 proportion; in addition, the layers (except for the top aqueous layer) contained 5% FCS. The second layer contained between 5 and 13 × 10⁴ developing amoebae or neutrophils (the lower number of cells was used for low magnifications).

For fluorescence microscopy (in which the 60× objective had a 250-μm working distance) the cAR1 strain was suspended in 15% Ficoll, placed directly in the base of the chamber, and overlayered with 12%, 5%, and 0% Ficoll and finally with mineral oil.

Needles.

A major problem arose in delivering the chemoattractant through a needle. Conventionally, delivery is done with a glass needle with an opening of around 1 μ; the chemoattractant held in the needle is pushed out using a modest pressure. However, we found that the device holding the needle vibrated too much, stirring the fluid in the chamber. The needle therefore was fixed to the chamber (as described above), allowing gravity to provide the pressure to release chemoattractant (with about 1 cm of chemoattractant solution above the level in the needle caused by surface tension alone). Several experiments were done in this way, but, especially with neutrophils, an additional problem arose. The first cell or two reaching the needle tip would sit on the end, and, it appeared, insert a pseudopod into it, thereby, we assume, blocking the tip. In any event, chemotaxis would cease. Using needles with wider openings simply resulted in turbulence around the needle tip caused by fluid flowing out of the tip or being sucked into the needle by capillary action.

Our final choice of needle was as follows: Glass tubing having an internal fiber to assist loading (part no. 30–0045; id 0.69mm; Harvard Apparatus) was pulled out (model P-2000; Sutter Instruments) to provide a needle about 6 cm long with a sealed end. This needle was loaded in a humidified chamber at 50 °C with about 1.5 μL 0.5% agarose (cat. no. 50071; Cambrex) in 1 mM EDTA so that an agarose plug about 2–3 mm long was formed in the needle tip when the capillary was cooled. Then 1 mM EDTA was added to the needle (to prevent drying), and the needle was stored in a humidified chamber. Before use, the tip was broken under the microscope to yield a blunt end with an opening of around 10–20 μm. The tip was immersed in, and the needle filled with, the chemoattractant solution (100 μM cAMP in KK2 buffer or FMLP in RPMI medium) and was held in that position for 24 h to allow the chemoattractant to diffuse throughout the agar plug. The “point” source of chemoattractant was produced by diffusion of the chemical out of the tip.

Microscopy.

Cells were observed in white light (exposure 10 ms) on an inverted Zeiss Axiovert S-100microscope with 5× or 10× objectives; time-lapse films (Movie S1 and Movie S4) were taken at 20- or 30-s intervals for about 1–3 h. At the end of recording, a z-stack (Movie S2) centered on the tip end might be taken. In all cases, imaging took place about 1.0–1.5 mm from the glass coverslip on the bottom of the chamber.

For observing the contours of a fluorescent cell, a Perkin-Elmer Ultraview spinning disk confocal microscope with a 60× water objective was used to collect a series of z-stacks (in 3-μm steps from −9 to +9 μm; exposure 0.5 s/frame) every 10 s (Movie S3). The needle tip was placed about 100 μm above the coverslip. Images were handled in Volocity (Perkin Elmer) and Image J software. All experiments were carried out at about 23–24 °C.