In vegetative cells, dynacortin localizes to the cell cortex and is especially enriched in protrusions such as pseudopodia, filopodia, lamellipodia, macropinocytic crowns, and cell-substrate structures called feet or eupodia. When cells change direction, green fluorescent protein fused with dynacortin (GFP-dynacortin) redistributes to the leading edge of the cell 89. This dynamic redistribution of GFP-dynacortin led us to speculate that dynacortin may have a role in chemotaxis. To determine the subcellular localization of dynacortin during chemotaxis, we expressed GFP-dynacortin in wild-type cells. We then imaged GFP-dynacortin distribution during the movement of aggregation-competent cells toward cAMP released from a micropipette. Epifluorescence microscopy revealed that GFP-dynacortin was diffusely localized throughout the cortex with occasional increases in concentration at the leading edge (Figure 1A; Additional file 1). The linescan in Figure 1B demonstrates dynacortin enrichment at the front of the migrating cell.

To obtain higher resolution images, we examined the distribution of dynacortin in the cell surface using total internal reflection fluorescence (TIRF) microscopy (Figure 1C,E; Additional files 2, 3). TIRF imaging allows the cortical layer to be imaged, greatly reducing the signal contribution from soluble GFP-dynacortin in the cytosol (Figure 1D). Dynacortin is organized into fibrous, punctuate structures, which are the actin-rich network near the cell surface that make up the actin feet 91213. Thus, the actin cross-linker dynacortin is recruited to highly dynamic regions of the cytoskeleton during chemotaxis.

Given the localization of dynacortin during chemotaxis, we tested the chemotaxis ability of cells depleted of dynacortin. Wild-type control cells were able to chemotax with high efficiency and assumed a highly polarized shape in response to cAMP (Figure 2; Additional files 4, 5). Here, we define cell polarization based on cell morphology in which the cells become highly elongated (pilate ellipsoid), and we use quantitative metrics (roundness) to assess cell shape (below). Cells that expressed a dynacortin hairpin plasmid had no detectable dynacortin and had a quantitative defect in cell polarization in response to cAMP (Figure 2; Table 1; Additional files 6, 7) 10. Though we frequently observed that dynacortin-deficient cells had highly impaired motility, failing to become polarized (for quantification see Table 1), the degree of immobility was highly variable and wild-type cells also had some highly rounded cells on occasion. Thus, we focused our quantitative analysis on motile cells from wild-type and dynacortin-deficient cells and measured the velocity, chemotactic index, directional persistence of the movement and the roundness (Table 1; see Methods for definitions of the parameters). Of the motile cells, dynacortin-depleted cells had statistically indistinguishable velocity, chemotactic index and directional persistence as compared to wild-type cells (Table 1). However, dynacortin-deficient cells (roundness coefficient: 0.6; Equation 1; Table 1) were significantly more round than wild-type cells (roundness coefficient: 0.4; Table 1). Thus, dynacortin is required for the cell to fully polarize in response to cAMP. We observed similar trends at 5.5 and 7 h of starvation, indicating that the defect in polarization is independent of developmental stage; significant increase in polarization is observed during this time period in wild-type cells 1415. Further, even though the dynacortin-deficient cells did not polarize as fully as wild-type control cells, GFP-myosin-II partially enriched in the rear half of the polarized, but not non-polarized, dynacortin-deficient cells (Figure 2G–I).

To quantify the motility defect further, we used a skeleton representation (see Methods) to characterize the dynamic changes in cellular morphology during chemotaxis (Figure 3). The skeletons of Ax2 control cells consisted primarily of a single segment indicating a lack of lateral protrusions (Figure 3A, Additional file 8). With the skeleton representation, we could separate the extension (red tips) and retraction (blue tips) of cell surface protrusions. The extending tips were relatively long-lived (Figure 3B) and the angles (mean ± SD: 1.2° ± 30.6°; Equation 2) of the extending tips co-aligned with the chemoattractant gradient (Figure 3C). In contrast, the skeletons of Ax2:dynhp cells showed multiple branches characteristic of cells with multiple protrusions (Figure 3D, Additional file 9). These projections occurred at a broader range of angles relative to the gradient and had a shorter duration (Figure 3E). Moreover, though these projections were also in alignment with the chemoattractant gradient on average, a broader distribution of extending protrusion angles was observed (mean ± SD: -8.0° ± 65.9°) (Figure 3F). Similarly, Ax3 control (mean ± SD: -2.7° ± 52.8°) and Ax3:dynhp cells (mean ± SD: -5.5° ± 74.6°) also produced protrusions with a similar angle relative to the gradient but Ax3:dynhp cells extended the protrusions with a larger variation in angles (Figure 3G,H). Thus, without dynacortin, cells generate more protrusions over a much broader distribution of angles than wild-type cells do.

The defect in polarization of dynacortin-deficient cells is similar to that observed for LY29004 (an inhibitor of PI3 kinase activity) treatment 1415. Therefore, we investigated the effect of LY29004 treatment of GFP-dynacortin localization (Figure 4). Using 5.5-h cells, wild-type cells expressing GFP-dynacortin were elongated with GFP-dynacortin enriched at the leading edge of the cell (Figure 4A). Within about 30 s of treating the cells with 50-μM LY29004, the cells nearly completely rounded up as described previously (Figure 4B) 14. However, by bringing the needle containing 1 μM cAMP close to the rounded cell, the cell accumulated GFP-dynacortin on the side facing the needle and extended a pseudopod within 20 s (Figure 4B). Within 2–3 min, the cells adapted to the LY29004 and began to chemotax towards the needle but never fully repolarized (Figure 4C). Significantly, GFP-dynacortin accumulated at the front end of the chemotacting LY29004-treated cells, suggesting that dynacortin accumulation at the cell front may be independent of PI3 kinase activity. We also treated dynacortin-deficient cells with LY29004; however, while the cells again showed lower overall polarization relative to wild-type cells, they were still able to chemotax towards the needle (data not shown).

For a population assessment, we examined the development on DB-agar. The dynacortin-depleted cells appeared largely indistinguishable from the wild-type control cells, forming normal looking fruiting bodies. However, at decreasing densities, the wild-type control cells repeatedly formed large, extensive streams whereas the dynacortin-depleted cells seldom formed streams, and when they did they were smaller than the wild-type control. Thus, in this population assay, dynacortin is required for development at low cell densities, consistent with the overall polarization and motility defects observed in the needle assays.

As dynacortin localizes to actin-rich dynamic cortical domains, we determined whether dynacortin contributes to the bulk actin polymerization dynamics in the cell. We compared the kinetics of in vivo bulk actin polymerization of dynacortin-silenced cells (wt: dynhp) to control wild-type cells in response to a chemotactic stimulus (Figure 5). Within 6 s of stimulation with 1-μM cAMP, wild-type and dynacortin-depleted cells showed a transient assembly of actin (~two fold), and then a decrease to basal levels by 30 s, followed by a smaller second (~1.2-fold) peak at 90 s. Thus, dynacortin does not have a detectable effect on the bulk actin assembly dynamics in response to cAMP. Further, basal actin levels were similar between wild-type and dynacortin-depleted cells (data not shown).

Because dynacortin enriches at the leading edge of chemotaxing cells and these motile cells are known to assemble new filamentous actin when stimulated with cAMP, we wondered whether dynacortin has an effect on actin assembly. To do this, we investigated the effect of dynacortin on actin assembly in vitro using pyrene-actin assays. In these experiments, the fluorescence of pyrene-actin increases as it assembles into filaments. Traditionally, these assays are performed in high-salt F-buffer, which contains 50 mM KCl and 1 mM MgCl₂. However, while performing some control experiments, we discovered that dynacortin drove actin assembly in G-buffer, which contains only 10 mM Tris-HCl and 0.2 mM CaCl₂. Because G-actin normally does not assemble under these low-salt G-buffer conditions, this dynacortin-mediated assembly represents a nearly infinite rate enhancement. Therefore, we tested dynacortin's role in actin assembly under both G- and F-buffer conditions and report these findings here.

We first measured the critical concentration of G-actin required for polymerization in the presence and absence of dynacortin. The critical concentration of G-actin assembled by 2.5 μM dynacortin under G-buffer conditions was only 0.03 μM, which is significantly lower than the 0.17 μM critical concentration for actin alone assembled by the addition of F-buffer (Table 2). Because the addition of dynacortin to F-actin caused a partial quenching of pyrene fluorescence, the critical concentration of actin under F-salt conditions could not be determined accurately (Table 2).

Then, we investigated the impact of dynacortin on the kinetics of actin polymerization under G-buffer conditions. We mixed increasing amounts of dynacortin with a constant amount of pyrene-labeled Ca²⁺-actin (10 μM) and the rate of polymerization was monitored over time by fluorescence intensity measurements. At dynacortin concentrations above 2 μM (actin:dynacortin ratio of 5:1), the G-actin assembled into polymers, and dynacortin increased the steady state amount of assembled actin in a dynacortin-concentration-dependent manner (Figure 6A,B). Because of the absence of an apparent lag phase (Figure 6A), the observed rate of polymerization could be obtained by fitting the curves to an exponential rise equation (Equation 3). The rate of polymerization increased linearly with increasing dynacortin concentration until a maximal value (k⁺ = 0.004 s^-1) was reached at 5-μM dynacortin. Further increase in dynacortin concentration resulted in no increase in rate beyond the maximal value. Considering the k⁺ and the slope of the concentration-dependent phase, the K_M was calculated to be 3 μM, which is similar to the measured 1.7 μM critical concentration for dynacortin to assemble polymers in G-buffer (Table 2). These results suggest that under low ionic strength conditions, dynacortin scaffolds actin monomers into filaments. Further, the absence of an apparent lag phase suggests that the mechanism of dynacortin-mediated actin assembly is different than the conventional nucleation-elongation mechanism of actin assembly.

Next, we verified that the increase in fluorescence was not due to an effect on the fluorescent probe but that dynacortin indeed bound actin under these low-salt conditions, using a co-sedimentation assay (Figure 6C). The solutions from the pyrene assay were centrifuged at high-speed centrifugation to pellet all the assembled actin and bound dynacortin. In the case of G-actin alone, no actin precipitated; most of the actin assembled using F-buffer precipitated at this speed. In the presence of G-actin and dynacortin, dynacortin co-precipitated with actin and had a saturation stoichiometry of one Dyn₂ dimer per actin monomer. This saturation stoichiometry is identical to the saturation stoichiometry for dynacortin-mediated actin bundling 910. More importantly, the amount of actin that assembled and pelleted increased as the dynacortin concentration increased. To independently confirm the presence of actin filaments, we examined the samples under the electron microscope and observed bundles of actin filaments in the presence of dynacortin under low salt conditions (Figure 6D). Staining actin filaments with rhodamine-phalloidin and visualizing them under the fluorescence microscope also revealed the presence of bundles (data not shown). Taken together, these data demonstrate that dynacortin drives assembly of G-actin into polymers under conditions that disfavor assembly and that the amount of polymer formed is proportional to the amount of dynacortin available. Thus, dynacortin acts as a scaffold or molecular staple that holds the actin monomers together under these low salt G-buffer conditions.

We also tested whether dynacortin can nucleate actin under standard F-buffer conditions. Under these conditions, the conventional lag phase associated with the nucleation-elongation reaction is observable. As the dynacortin concentration was increased, the lag phase for actin assembly decreased slightly (Figure 6E). The lag phase is derived from the time required for actin nuclei to form. To analyze this more quantitatively, we determined the time for the actin fluorescence signal to increase to 10% of the maximum signal, indicating that the actin is 10% assembled (Equation 4). The reciprocal of the time to 10% has been demonstrated to be a close approximation of the nucleation rate 16. Dynacortin weakly increased the apparent nucleation rate in a dynacortin-concentration-dependent manner (Figure 6F). Overall, we conclude that dynacortin's primary role is in actin filament cross-linking with a possible effect on filament stability.

Dictyostelium discoideum Ax3(Rep orf+) 8, Ax2, or myoII::GFPmyoII:pDRH, HygR, RFP-α-tubulin:pDxA-Bl cells were cultivated in Hans' enriched HL-5 medium in polystyrene Petri dishes at 22°C 8. Wild-type (wt) control and wt: dynhp cells were constructed as previously described 10. Control cells contain the empty vector, pLD1A15SN. Western analysis using anti-dynacortin antibodies was used to detect dynacortin levels in control and wt: dynhp cells 810.

For development, cells were washed in developmental buffer (DB; 10 mM phosphate buffer, 2 mM MgSO₄, 0.2 mM CaCl₂) and plated on DB agar at known densities (cells/cm²). Twofold dilution series from 1 × 10⁶ to 6 × 10⁴ cells/cm² were performed three times in duplicate (total n = 6 for each genotype). Development was monitored for up to 30 h.

To starve cells for chemotaxis assays, cells were washed in DB buffer, resuspended at 2 × 10⁷ cells/mL in 5 ml DB and rotated at 100 rpm for 1 h. Cells were then pulsed with 50-nM cAMP at 6 min intervals for 4–6.5 h. For chemotaxis assays, 1-μM cAMP was released from a microcapillary pipette (0.5 μm inner diameter) with 40 hPa of pressure. Cells were imaged using an Olympus IX81 motorized microscope using a 40 × (numerical aperture (NA) 1.3) objective and a 1.0× optivar. Images were collected every 5 s for 20 min using Metamorph imaging software (Molecular Devices Corp., Downingtown, PA USA).

To quantify chemotaxis, we first determined the fraction of cells that were motile. To do this, we followed all of the cells that were present in the initial field of view; this eliminates the impact of cells that migrate into the field, which would lead to an overestimation of the percentage of motile cells. To analyze the motile cells, we determined the velocity 33, chemotactic index (cosΘ) 27, directional persistence 34, and roundness 33. The fraction of chemotactic cells was determined by following the behavior of every cell in the initial microscopic field over a 20 min video recording. For the detailed analysis of motile cells, we only followed cells that did not collide with another cell during the movie to avoid complications from obstacles. Analysis was performed using a Matlab Image Processing Toolbox (Mathworks, Natick, MA, USA). Individual cells were first segmented and tracked during the course of each video. The centroid was then monitored at 15-s time resolution for the duration of the video. Similarly, we analyzed cell motility by monitoring the centroid position of each cell at 2-min time resolution using Image J 35. The automated (Matlab) and manual (Image J) methods gave statistically identical results. Velocity reported was the mean velocity of each cell as measured using 15-s time windows. Directional persistence is the net path taken by the cell over the entire video divided by the total path taken using a 15-s time resolution. The chemotactic index (cosΘ) is the cosine of the angle defined by the cell position at time t_+15s to the cell position at time t and from the cell position at time t to the needle; thus, the initial position of the cell at time t is the vertex of the angle. Roundness (R) is calculated as:

R = 4πA/L²

Where A is the area of the cell and L is the perimeter length. By this calculation, a circular cell will have an R-value of 1 and a line segment will have an R-value of 0. Each cell had a mean velocity, cosΘ, directional persistence, and roundness, which were averaged for all cells within the genotype. Standard errors of the mean were calculated from the distribution of means for each genotype. Values are presented as mean ± standard error of the mean (SEM). Student t tests were used to evaluate statistical significance of differences between control and dynacortin-depleted strains for the various parameters.

Image analysis of pseudopod formation dynamics was based on a skeleton representation of cell shape 36 and the detection of shape changes between successive frames, using the Matlab Image Processing Toolbox (Mathworks). First, cells were segmented and tracked. For each frame, the boundary of the segmented image was smoothed by curve fitting using recursive least-square estimation 37. For each frame, the skeleton of each shape was computed. These were pruned in two steps. To avoid spurious branches, a length threshold (8 pixels of length – approximately 2.46 μm) was set in the first step. Single branches whose length did not exceed the threshold were eliminated; multiple branches with comparable lengths that were shorter than the threshold were combined from their common root by averaging their terminal locations. The second pruning step takes into account the changes over time of the cell shape. To this end, cell shape differences between the current and previous frame were detected. Skeleton tips were then tested to see whether they were close to the areas with cell shape differences. Those farther than 1.8 μm were eliminated. Also deleted were the tips that did not point towards areas in which the cell shape changed. Finally, branches were classified as extensions or retractions depending on whether they pointed towards regions in which the cell shape was expanding or shrinking. Tip angles are defined as the angle between the rays emanating from the cell's centroid to the needle location and extension/retraction tips. To determine the time course of the tips, differences in tip angle between up to four successive frames were computed and compared to an arbitrary threshold of 60°.

Angle distributions for the extending tips were fitted by a Gaussian with finite support:

c σ 2 π e − ( θ − μ ) 2 / 2 σ 2 , θ ∈ [ − 180 ∘ , 180 ∘ ) MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGacaGaaiaabeqaaeqabiWaaaGcbaqbaeqabeGaaaqaaKqbaoaalaaabaGaem4yamgabaacciGae83Wdm3aaOaaaeaacqaIYaGmcqWFapaCaeqaaaaakiabdwgaLnaaCaaaleqabaGaeyOeI0IaeiikaGIae8hUdeNaeyOeI0Iae8hVd0MaeiykaKYaaWbaaWqabeaacqaIYaGmaaWccqGGVaWlcqaIYaGmcqWFdpWCdaahaaadbeqaaiabikdaYaaaaaGccqGGSaalaeaacqWF4oqCcqGHiiIZcqGGBbWwcqGHsislcqaIXaqmcqaI4aaocqaIWaamdaahaaWcbeqaaiablIHiVbaakiabcYcaSiabigdaXiabiIda4iabicdaWmaaCaaaleqabaGaeSigI8gaaOGaeiykaKcaaaaa@5196@

The parameters μ and σ were obtained by optimization using the simplex search method in Matlab.

Wild-type GPP-dynacortin (wt: GFP-dyn) cells were imaged on an Olympus IX81 epifluorescence/total internal reflection fluorescence (TIRF) microscope. Cells were imaged in DB buffer. For epifluorescence imaging, cells were illuminated with a Xenon lamp, and a 40 ×, 1.3 NA oil-immersion objective with a 1.6× optivar were used for image collection. For TIRF imaging, cells were illuminated with a 488-nm laser, and a 60 × 1.45 NA objective and 1.6× optivar were used for image collection. Images were collected using Metamorph Software (Molecular Devices) and processed using Image J 35 and Adobe Photoshop (Adobe Systems, Inc., San Jose, CA USA). For GFP-myoII imaging, myoII (mhcA; HS1 38) cells complemented with GFP-myoII: pDRH 39 and expressing either the empty pLD1A15SN (control) vector or pLD1A15SN:dynhp 10 were used.

The relative filamentous actin content was determined from tetramethylrhodamine B isothiocyanate-phalloidin (TRITC-phalloidin) staining of Dictyostelium cells 27. Briefly, developed cells were resuspended at 3 × 10⁷ cells/mL with 3 mM caffeine in PM buffer (10 mM phosphate buffer, 2 mM MgSO₄) and shaken at 200 rpm for 15–30 min. At various times after 1-μM cAMP addition, 100-μL aliquots were collected, fixed and stained with TRITC-phalloidin. F-actin was proportional to the amount of phalloidin fluorescence (excited at 540 nm and emitted at 570 nm) extracted from cell pellets.

Chicken skeletal muscle actin was purified from acetone powder 40, gel filtered and labeled with pyrenyl-maleimide 41. Recombinant dynacortin was purified following the method described previously 10.

Dynacortin and pyrene-labeled actin (10% and 60% pyrene-labeled) in G-buffer (10 mM Tris-HCl, pH 7.5, 0.2 mM CaCl₂, 0.2 mM ATP, 5 mM DTT) was precleared at 100 000 g for 1 h before use. To assess the effect of dynacortin on actin polymerization, 10-μM pyrene-labeled actin was mixed with varying concentration of dynacortin under G-buffer conditions. No difference was observed in dynacortin-mediated G-actin assembly in G-buffer conditions between 10% and 60% pyrene-labeled actin. Changes in fluorescence were monitored by exciting at 365 nm and collecting emission at 385 nm. The rate of polymerization, monitored by the change in fluorescence signal, was calculated using the relationship:

F_t = F_max - F_initial·e^-kt

Where F_t is the fluorescence amplitude at any given time t, F_max is the end point, and F_initial is the starting point of the fluorescence amplitude, and k is the polymerization rate constant determined from fitting the polymerization curve to the above equation.

For nucleation in F-buffer conditions, varying concentrations of dynacortin were added to 2.5 μM G-actin (10% pyrene-labeled) in F-buffer (G-buffer plus 2 mM MgCl₂ and 50 mM KCl). Nucleation rates were calculated by determining the time required to achieve 10% polymerization (0.1(F_max-F_initial)) as described in 16:

Rate_nucleation = 1/(t_{(0.1(Fmax-Finitial))}).

This is only an approximation of the nucleation rate as nucleation involves multiple steps.

The samples from the fluorescence assay were centrifuged for 30 min at 100 000 g in a TLA 100 rotor. Only polymerized actin sediments under these conditions. The pellets were resuspended in 1 × Laemmli's sample buffer and equivalent amounts were loaded on 15% SDS-PAGE gels. Densitometry of Coomassie Blue stained gels was used to determine the fraction of actin that was polymerized by dynacortin.

10-μM G-actin (10% pyrene-labeled) was polymerized by addition of 2 mM MgCl₂ and 50 mM KCl 25°C for 1 h. F-actin was diluted to the desired concentration in F-buffer. The pyrene fluorescence of F-actin at each concentration was monitored as described above. Unassembled G-actin pyrene fluorescence at each concentration was also measured to determine the background fluorescence. To determine the critical concentration of actin or dynacortin, the net change in pyrene fluorescence of an actin polymerization reaction was measured as a function of increasing actin or dynacortin concentration. Critical concentrations were obtained from the linear fits of the data, which were used to determine the x-intersect of the comparator unassembled and assembled samples.

Samples containing actin and dynacortin were deposited on carbon-coated Formvar grids, negatively stained with 1.5% uranyl acetate, and observed using a Phillips CM-120 TEM, operating at 80 kV.