• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of biophysjLink to Publisher's site
Biophys J. Sep 1, 2007; 93(5): 1787–1796.
Published online May 18, 2007. doi:  10.1529/biophysj.107.104356
PMCID: PMC1948060

Biased Random Walk by Stochastic Fluctuations of Chemoattractant-Receptor Interactions at the Lower Limit of Detection

Abstract

Binding of ligand to its receptor is a stochastic process that exhibits fluctuations in time and space. In chemotaxis, this leads to a noisy input signal. Therefore, in a gradient of chemoattractant, the cell may occasionally experience a “wrong” gradient of occupied receptors. We obtained a simple equation for Ppos, the probability that half of the cell closest to the source of chemoattractant has higher receptor occupancy than the opposite half of the cell. Ppos depends on four factors, the gradient property equation M1 the receptor characteristic equation M2 a time-averaging constant I, and nonreceptor noise σB. We measured chemotaxis of Dictyostelium cells to known shallow gradients of cAMP and obtained direct estimates for these constants. Furthermore, we observed that in shallow gradients, the measured chemotaxis index is correlated with Ppos, which suggests that chemotaxis in shallow gradients is a pure biased random walk. From the observed chemotaxis and derived time-averaging constant, we deduce that the gradient transducing second messenger has a lifetime of 2–8 s and a diffusion rate constant of ~1 μm2/s. Potential candidates for such second messengers are discussed.

INTRODUCTION

Chemotaxis is a pivotal response of many cell types to external spatial cues. It plays important roles in diverse functions such as finding nutrients in prokaryotes, forming multicellular structures in protozoa, tracking bacterial infections in neutrophils, and organizing the embryonic cells in metazoa (13). Chemotaxis is achieved by coupling gradient sensing to basic cell movement. Whereas prokaryotes are too small (~1 μm) to sense spatial gradients and therefore rely on temporal changes of the chemoattractant concentration, eukaryotes are large enough (10–20 μm) to process spatial gradients as well as temporal gradients. Prokaryotes adjust their tumbling frequency in response to temporal changes of the chemoattractant concentration (4,5). In eukaryotes, the difference in receptor occupation across the cell body leads to pseudopod extension directed toward the higher concentration in a chemical gradient (6,7).

In Dictyostelium cells, extracellular cAMP functions as chemoattractant that is detected by specific G-protein coupled surface receptors. Cells are able to sense the cAMP gradient over a wide range of concentrations from 0.1 nM to 10 μM, with concentration differences of only 1–5% across the cell (8,9). A significant chemotactic response is induced at a threshold close to 0.5 nM cAMP with a gradient of 9%. Dictyostelium cells have ~40,000 cAMP receptors with a dissociation constant (KD) that is ~100 nM, implying that at threshold concentrations, ~200 receptors are occupied in total. Since the cells are ~10 μm in size and receptors are distributed homogeneously around the cell (10), at the threshold gradient the front half of the cell has ~5 more occupied receptors than the back half of the cell. Binding of cAMP to the receptor is a stochastic process and therefore fluctuates in time, and consequently will vary in space as well. At these low threshold concentrations, the receptor occupancy statistics can be approximated by a Poisson distribution (11). The standard deviation in receptor occupancy then equals the square root of the mean receptor occupancy. Thus at threshold chemotaxis with an average occupancy of ~200 receptors/cell, the fluctuation around the mean is ~14 receptors/cell, whereas the difference between front and back half of the cell is only ~5 receptors/cell. Such fluctuations of input signals have been observed by single molecule imaging of cAMP-binding to living Dictyostelium cells (11). How does the chemotactic system obtain reliable information regarding the cAMP gradient with these noisy inputs? This question is critical, because the downstream gain present in chemotactic signaling will amplify small changes in cAMP concentration, and such systems will therefore amplify the noise as well.

Berg and Purcell (12), and more recently Bialek and Setayeshgar (13), have shown that statistical fluctuations limit the precision by which a cell, in a given time T, can determine the concentration of a chemoattractant in the surrounding medium. They have shown that the transduction time acts as time averaging the input signal, which will reduce transducer noise. For leukocytes, a stochastic model has been proposed based on the small fluctuations in turning direction at the front of the cell, which has an intrinsic polarity axis (1416). However, Dictyostelium and many other amoeboid cells are not polarized in buffer. At these conditions, the direction in which a new pseudopod is extended is only slightly correlated with the direction of a previous pseudopod, and cell tracks essentially have the characteristics of a random walk (1719). In a gradient of chemoattractant, pseudopodia can still be formed in all directions, but it occurs more frequently in the direction of the gradient. In such a biased random walk, noise is an intrinsic property of chemoreception.

We have analyzed Dictyostelium chemotaxis in shallow cAMP gradients where noise is a more dominant component of the signal transduction cascade, and derived an equation for the probability that the cell experience a higher receptor occupancy at the side of the cell closest to the chemoattractant source than at the opposite side of the cell. This probability is called Ppos, and includes experimental estimates of the noise and time-averaging period. We observed that the measured chemotaxis index is directly proportional to Ppos, suggesting that chemotaxis in shallow gradients is a pure biased random walk.

MATERIALS AND METHODS

Strains and growth conditions

The wild-type Dictyostelium discoideum strain AX3 was used in all experiments. Cells were grown in shaking culture in HG5 medium (contains per liter: 14.3 g oxoid peptone, 7.15 g bacto yeast extract, 1.36 g Na2HPO4·12H2O, 0.49 g KH2PO4, 10.0 g glucose) at a density between 5 × 105 and 6 × 106 cells/ml. Cells were harvested by centrifugation for 3 min at 300 g, washed in phosphate buffer (PB) (10 mM KH2PO4/Na2HPO4, pH 6.5) and starved in PB in 6-well plates (Nunc) for 5 h. Cells were then resuspended in PB, centrifuged and washed once in PB, and resuspended in PB at a density of 5 × 106 cells/ml.

Chemotaxis

Three assays were used to quantify chemotaxis. In the pipette assay, a droplet of ~0.5 ml of the cell suspension was placed on a microscope glass slide. Cells were allowed to adhere to that glass for 15 min, and then cells were stimulated with a micropipette. The Eppendorf femtotip, filled with different concentrations of cAMP, was placed just above the glass surface and cAMP was released at a pressure of 25 hPa. Cells were monitored by phase contrast microscopy and images were captured every 10 s during 20 min. The formation of the cAMP gradient was deduced from experiments in which the fluorescent dye Lucifer yellow was mixed with cAMP in the pipette. The fluorescence intensity at different distances of the pipette was recorded using a confocal fluorescent microscope in pixel elements (0.404 × 0.404 μm), and calibrated using the fluorescence intensity of a diluted Lucifer Yellow added homogeneously to the bath. The results (data not shown) reveal that the gradient is formed very fast (the time at which half-maximal equilibrium is reached was ~10–20 s), and follows Eq. A1 at distances from the pipette beyond 15 μm. From these experiments, we obtained the portionality constant of Eq. A1, α = 0.05.

The second chemotaxis assay employs a modified Zigmond chamber (20). On a microscope slide, a glass bridge of 24 × 2 mm was places on top of two supporting glass strips with thickness 0.15 mm. Cells were places under the bridge. A block of agar containing only PB buffer was placed at one side of the bridge, while a block of agar containing cAMP and buffer was placed at the other side. A gradient of cAMP was formed under the bridge. Cells were observed by phase contrast microscopy in an area of 350 × 270 μm at a distance of 700 μm from the agar block containing cAMP, and images were captured every 10 s during 20 min. The dye bromophenol blue was used to follow the formation of the cAMP gradient (20).

The small population assay was used as the third chemotaxis assay (8,21). Droplets of ~0.1 μl of 5-h starved cells (6 × 106 cells/ml) were placed on nonnutrient hydrophobic agar (11 mMKH2PO4, 2.8 mM Na2HPO4, 7 g/l hydrophobic agar). Chemotaxis toward cAMP was tested by placing a second 0.1 μl droplet, with the indicated concentration of cAMP, next to the droplet of cells. The distance between the droplets, defined by the distance between the center of the cAMP-containing droplet and the closest edge of the cell-containing droplet, was either 250 or 650 μm. The distribution of the cells in the droplet was observed, and scored positive when at least twice as many cells were pressed against the side of the higher cAMP concentration as against the other side of the droplet. Recorded was the fraction of droplets scored positive.

The motile behavior of cells in spatial gradients of cAMP was analyzed using computer-assisted methods previously described (22). Briefly, images of the pipette assay and the modified Zigmond assay were used to determine the contour of the cell and the position of the cell centroid. Chemotactic index, defined as the distance moved in the direction of the cAMP gradient divided by the total distance moved, was computed from the centroid positions for each subsequent image, and averaged for each cell over at least 20 images. Reported were the means and standard error of the means of at least 30 cells from two or three experiments.

RESULTS

Chemotaxis at different gradients of cAMP

Three different assays were used to measure the chemotactic response of Dictyostelium cells: a micropipette assay, a modified Zigmond chamber, and the small population assay. The equation for the formation of cAMP gradients around the cell are presented in the Appendix We measured the chemotaxis index (Ψ) toward a pipette with buffer and obtained Ψ = 0.01 ± 0.05 based on n = 24 (mean ± SE), which is statistically equal to zero. The threshold gradient for chemotaxis is defined as the smallest gradient that yields a chemotaxis index that is statistically significant above this value in buffer. With n = 24 and P < 0.01, the threshold chemotaxis index is close to Ψ = 0.2.

We have determined the chemotaxis index of cells at different distances from the pipette, demonstrating that cells exhibit significant chemotaxis as far as 700 μm from a pipette filled with 10−4 M cAMP (Fig. 1 A). A half-maximal chemotaxis response (Ψ = 0.5) is observed at ~300 μm from the pipette. When the pipette contains lower concentrations of cAMP, cells far away from the pipette no longer respond, but closer to the pipette cells still exhibit significant chemotaxis. In the modified Zigmond chamber, cells are deposited under a 2 mm wide glass bridge. A cAMP gradient is formed by placing agar blocks with buffer on one side of the bridge and an agar block with cAMP on the other side (20). Cells are observed at a distance of ~700 μm from the source. Using 1 μM cAMP in the source block, the gradient is 0.5 nM/μm at a mean cAMP concentration of 660 nM, yielding a concentration difference of ~0.7% over the cell length. This gradient induced a chemotaxis index of ~0.7 (Fig. 1 B). The chemotaxis index becomes smaller when the cAMP concentration in the source block is reduced. At 10 nM cAMP, the threshold chemotaxis index of ~0.2 is observed, whereas the chemotaxis index is statistically indistinguishable from zero at a concentration of 1 nM cAMP. The small population assay scores the chemotactic response of a population of cells toward a cAMP source that has been applied at some distance away as a single dose. When the cAMP and Dictyostelium droplets are placed close together (x = 250 μm), significant chemotaxis is detected at 3–10 nM cAMP in the droplet. A half-maximal response is induced by 30 nM cAMP, whereas the response saturates at 1 μM cAMP in the droplet (Fig. 1 B). When the distance between cAMP and cells is increased to x = 650 μm, ~100-fold higher cAMP concentrations are required to induce a chemotactic response with similar strength as with x = 250 μm.

FIGURE 1
Chemotactic activity of Dictyostelium cells toward cAMP. (A) Cells were stimulated with micropipettes containing different concentrations of cAMP as indicated. (B) Chemotaxis with the small population assay at different distances between cells and cAMP ...

Table 1 summarizes the experimental conditions of the assays where we observed a threshold chemotaxis of Ψ = 0.2. Equations A1A6 of the Appendix were used to calculate the concentration and spatial gradient around the cell, showing that a threshold chemotactic response can be induced at a 60-fold difference in cAMP concentrations and fourfold difference in spatial gradient, depending on the assay used. The number of occupied receptors per cell was calculated, as well as the difference of occupied receptors between the front and back half of the cell. Chemotaxis can be induced when the difference between the front and back of the cell is as little as five occupied receptors when only 0.3% of the receptors are occupied with cAMP. At higher cAMP concentrations, when ~7% of the receptors are occupied, the difference between the front and back is still only 10 occupied receptors. These experimental data clearly show that threshold chemotaxis can be obtained at different magnitudes of the spatial gradient and absolute concentration, which will be used below to derive estimates of the noise parameters for chemotaxis.Table 1

TABLE 1
Conditions for threshold chemotaxis (Ψ = 0.2)

Receptor noise leading to biased random walk at threshold chemotaxis

The cell measures the cAMP gradient by means of surface cAMP receptors. The input signal is a spatial cAMP gradient, which is detected by the cell as the difference of occupied receptors between the front and back half of the cell, ΔR*. In shallow gradients (<10% across the cell), ΔR* is given approximately (see the Appendix A7) by

equation M3
(1)

where ΔC is the cAMP concentration difference between the front and back of the cell, Rt is the number of receptors (40,000/cell), and KD is the dissociation constant of the receptor (100 nM).

The interaction between cAMP and its receptor has been investigated in vivo by single molecule detection, demonstrating that the interaction is stochastic (11). Random association and dissociation reactions of cAMP-receptor interactions inevitably leads to fluctuations in the number of occupied receptors, and consequently receptor activation is noisy. The noise in the chemotactic system due to fluctuations in receptors occupancy follows binomial statistics, depending on the number of occupied and unoccupied receptors. When the chemotactic system time averages the receptor occupancy over a characteristic sampling time T, the variance of the number of occupied receptors after sampling (equation M4) at low cAMP concentrations (C < KD) is given by (see the Appendix)

equation M5
(2)

where I is introduced as the sampling fold, which indicates how often a receptor measures the cAMP concentration in the time T. The sampling fold I is given by

equation M6
(3)

where at low cAMP concentrations equation M7 which denotes the lifetime of the cAMP-receptor complex.

Besides noise due to fluctuations in the cAMP-receptor interaction, the system will also contain nonreceptor noise, originating from fluctuations in the second messenger system downstream of the receptor. This nonreceptor noise σB will increase the variance and reduce the accuracy by which the spatial gradient can be detected.

The signal/noise ratio (SNR) is then given by

equation M8
(4)

Note that in the absence of nonreceptor noise (i.e., σB = 0) Eq. 4 reduces to

equation M9
(5)

Thus, SNR consists of four parts: nonreceptor noise σB, the cAMP gradient property equation M10 receptor property equation M11 and the sampling fold equation M12 The SNR increases, i.e., the gradient is measured more accurately, when nonreceptor noise σB is smaller, when the spatial cAMP gradient ΔC is steeper at lower average cAMP concentration C, when the number of receptors Rt or the affinity of the receptors for cAMP equation M13 increase, or when the sampling fold I increases.

Due to stochastic fluctuations of occupied receptors, the front of the cell may occasionally experience a “wrong” lower receptor occupancy than the back of the cell. The probability that the cell experiences a net positive signal in the direction of the gradient is called Ppos and is given by

equation M14
(6)

where Erf is the error function.

We are especially interested in chemotaxis at threshold gradients, where SNR and Ppos are expected to be small. The inset of Fig. 2 reveals that Ppos reaches significant values at rather low values of SNR. Thus, the probability that the front of the cell measures a positive gradient is 0.2 at a SNR of 0.15, i.e., when the noise is sevenfold larger than the signal. Furthermore, the figure shows that Ppos is linear with SNR for SNR < 0.7 with a slope of equation M15 i.e., equation M16 We calculated Ppos as function of the distance from a pipette with 10−4 M cAMP for different values of nonreceptor noise σB, sampling fold I, and receptor property Rt/KD. Fig. 2 shows that with σB = 0 and I = 1 (curve c), Ppos reaches half-maximal value of 0.5 at a distance of ~300 μm from the pipette. At a nonreceptor noise of 10 or 100 (curves a and b), half-maximal Ppos is obtained much closer to the pipette. In contrast, increasing the sampling fold I to 3 or 10 (curves d and e) leads to a significant increase of Ppos, by which cells farther away from the pipette may respond. Increasing the number of receptors Rt or the affinity of the receptor 1/KD has the same effect as increasing the sampling fold: the value of Ppos increases by which cells may detect the cAMP gradient more easily. The observed values for the chemotaxis index Ψ are also presented in Fig. 2, demonstrating that the experimental data exhibit the same distance-dependency as the calculated curves for Ppos. This is especially valid far away from the pipette. Closer to the pipette within a distance of ~150 μm, the gradient becomes very steep, and consequently Ppos approaches 1.0, whereas the chemotaxis index increases slowly from 0.8 to 0.9 but never reaches the value of 1. These results suggest that in shallow gradients, the input signal for chemotaxis is closely related to Ppos.

FIGURE 2
Measured chemotaxis index, and calculated Ppos at different distances from a micropipette containing 10−4 M cAMP. Ppos is the probability that the cell senses a gradient toward the pipette, and was calculated for different values of nonreceptor ...

The calculations presented in Fig. 2 reveal that different combinations of σB and I can explain the observed values of the chemotaxis index in the pipette assay. The chemotaxis data from the three different assays, exhibiting different gradient properties, were used to obtain experimental estimates for σB and I. For this, we assume Ψ = Ppos and consider only the conditions with SNR < 0.7, by which equation M17 Thus the assumptions are equation M18 and equation M19 and with these assumptions it holds that

equation M20
(7)

A plot of equation M21versus C will yield a straight line with slope equation M22 and intercept at the y axis of equation M23 The experimental data for threshold chemotaxis from Table 1 (Ψ = 0.2), and all experimental data with Ψ < 0.5 are presented in Fig. 3. The data are described accurately by a straight line. From the slope, we calculated the sampling fold, yielding I = 1.82 (range 1.24–2.40; 95% confidence limit). The nonreceptor noise was calculated from the intercept with the ordinate, yielding σB = 25.2 (range 15.6–34.8). In this figure, we also plotted the calculated curve for σB = 0, which results in a significant less accurate fit of the experimental data.

FIGURE 3
Plot of equation M24versus C to derive estimates for nonreceptor noise σB and sampling fold I. The chemotaxis data of Fig. 1 and Table 1, and the corresponding values of the cAMP concentration C and spatial gradient [nabla]C were used. (Solid symbols) Chemotaxis ...

Now that we have estimates of all parameters of SNR, we next investigated the relationship between the calculated Ppos and the chemotaxis index Ψ as measured in Fig. 1. Fig. 4 presents all observed chemotaxis data from the three assays as a function of the calculated Ppos using the appropriate equation for C and ΔC, all at I = 1.8 and σB = 25. The results show a remarkable close correlation between Ppos and Ψ for all assay conditions. The curve is linear with slope 1 at Ppos and Ψ below 0.5, and levels off to Ψ = 0.8 when Ppos approaches 1. The results strongly suggest that in shallow gradients, the chemotaxis index depends largely if not solely on Ppos, i.e., the probability that the half of the cell closer to the cAMP source receives a stronger receptor signal than the other half of the cell. Thus chemotaxis in shallow gradients is well described by a biased random walk.

FIGURE 4
Chemotaxis index as function of Ppos. The chemotaxis index presented in Fig. 1 was determined with three assays using different experimental conditions. Equations A1A6 were used to calculate the cAMP gradient around the cell in these assays, ...

DISCUSSION

Model for bias random walk

Dictyostelium cells in buffer extend a pseudopod on average every ~30 s; a pseudopod has a lifetime of ~60 s. We suppose that the time and place where a new pseudopod is made is regulated by activators and inhibitors in the cytoplasm. Alternatively, cells may extend a new pseudopod by splitting of an existing pseudopod, and one of the pseudopodia is stabilized and the other retracted (23), depending on the local balance between activators and inhibitors. We envision that cAMP receptor activation will influence the activity of one or multiple of these activators and inhibitors, thereby affecting the formation of new pseudopodia or stabilization of existing pseudopodia. In a shallow gradient of cAMP, this will result in a bias of the random formation of pseudopodia such that they are made more often in the direction of the gradient than in other directions. In such a probabilistic model for chemotaxis in very shallow gradients, there in no a priory need for gradient amplification steps.

Ppos, the probability to decipher a cAMP gradient

Dictyostelium cells are exquisitely sensitive to a spatial gradient of cAMP. At threshold chemotaxis (Ψ = 0.2), the minimal spatial gradient is ~5 pM/μm (Table 1), similar to what has been observed previously with the small population assay (3.6 pM/μm (8)) and microfluid devices (~3–5 pM/μm (24)). The difference in receptor occupancy between the two halves of the cell closest and farthest away from the cAMP source is only 5–10 occupied receptors per cell at threshold chemotaxis. We have analyzed chemotaxis in shallow gradients in the context of stochastic fluctuations of receptor occupancy, and derived an equation for Ppos, the probability that in a cAMP gradient the front half of the cell has higher receptor occupancy than the back half of the cell. We observed a very close correlation between the calculated Ppos values and the observed chemotaxis index. This close correlation was obtained with several different chemotaxis assays that depend either on the behavior of cell populations, or on the behavior of individual cells in stable spatial gradients or transient spatial gradients. These results suggest that the probability that a cell makes a new pseudopod or stabilizes an existing pseudopod in the direction of a shallow cAMP gradient depends on the probability that the cell has correctly determined the direction of the cAMP gradient.

Ppos depends on cAMP gradient, receptors, nonreceptor noise, and sampling time

The value of Ppos depends on four parameters: the cAMP gradient propertyequation M27 the receptor property equation M28 the nonreceptor noise σB, and the sampling fold I. The observation that chemotaxis at low cAMP concentrations depends on equation M29 strongly supports the hypothesis that chemotaxis in shallow gradients is mediated by stochastic fluctuations of activated receptors. The equation M30 in the denominator is derived from the receptor noise, whereas ΔC in the nominator is derived from the signal: At low cAMP concentrations, the difference in receptor occupancy between the front and back of the cell is proportional to the absolute cAMP gradient ΔC (Eq. 1). The dependence of chemotaxis on equation M31 has been observed before in experiments in which the chemotaxis threshold was measured with increasing added background cAMP concentrations (9). The experiments were interpreted in the context of adaptation of the receptor to increasing background concentrations and not in the context of stochastic fluctuations of receptor occupancy. That interpretation is probably correct, because the applied background cAMP concentrations were very high (10−7–10−5 M), essentially occupying all receptors unless the affinity of the receptors was changed by adaptation. These results suggest that chemotaxis depends on equation M32 both at low receptor occupancy, due to stochastic fluctuations of receptor occupancy, and at high receptor occupancy, due to adaptation mechanisms.

Ppos, and thus chemotaxis, is proportional to equation M33 The sampling fold I is the number of independent measurements of receptor occupancy, which depends on the sampling time T and the lifetime τR of the cAMP-receptor complex as described in Eq. 3. It is clear that chemotaxis will improve when the sampling fold I is large, which will be the case when the sampling time T is large or the lifetime τR of the cAMP-receptor complex is small. However, the consequence of a large sampling time T is that second messengers that transduce the signal will diffuse and lose spatial information. In addition, a very short lifetime τR of the cAMP-receptor complex will lead to a lower affinity of the receptor for cAMP, which reduces the sensitivity to detect low chemoattractant concentrations. Different kinetic forms of the receptor have been observed, for which the lifetimes τR of the active receptor have been obtained between 0.7 and 3.5 s (25,26). We derived the experimental value for sampling fold I = 1.8. From Eq. 3, we derive that T/τR = 2.1, and therefore the sampling time T of the receptor/second messenger system is in the range of 1.5–7 s. It has been observed that Dictyostelium cells can extend a pseudopod in the direction of a micropipette with cAMP within ~5–10 s after application of the cAMP gradient (2729). The importance of the sampling time for understanding the properties of the transducing second messenger will be discussed below.

The probability to detect a spatial gradient of chemoattractant is reduced by nonreceptor noise, σB. The source of nonreceptor noise can be a basal level of second messenger, fluctuations in the second messenger response, and fluctuations of the cAMP concentration in the vicinity of the receptor. We have not investigated the individual components that lead to nonreceptor noise. We obtained the experimental value for nonreceptor noise σB = 25. How does this value relate receptor noise σR? Equation 2 yields a receptor noise σR of 25 with I = 1.8 at R* = 347 occupied receptors/cell, which will occur at a cAMP concentration of ~1 nM. Thus, nonreceptor noise is very small, and above 1% occupancy of the receptors, the noise in the system is already dominated by receptor noise. This suggests that the second messenger system that mediates chemotaxis shows significant increases at very low cAMP concentrations, either because the system is fully activated at low receptor occupancy, or because basal levels of the second messenger are extremely low.

The transducing second messenger

We wish to deduce some general characteristics of a hypothetical second messenger M that transduces the signal from occupied chemoattractant receptors to activation of a pseudopodium in Dictyostelium and other systems. It is likely that the sampling time T is approximately equal to lifetime τM of the second messenger, i.e., τM = T. It is clear that the sampling fold I can be large when the lifetime of the second messenger τM is large. However, to transduce the spatial information of occupied receptors to a local pseudopod, the second messenger should not diffuse too far from its place of production. On the other hand, mixing of second messenger molecules in the vicinity of their place of synthesis will reduce the temporal and spatial noise of the system. A molecule M that diffuses through a medium where it is degraded will have an average dispersal length lM from the place of synthesis given by equation M34 where γ is a constant that depends on the dimensions (γ = 1, π/2, and 2 for diffusion in one, two, or three dimensions, respectively) and DM is the diffusion rate constant of M (30). Pseudopodia are protrusions of 2–4 μm in width. This suggests that the optimal dispersion length of M is between 1 and 3 μm. With these assumptions and γ = 1.5, it follows that equation M35 It is unlikely that the sampling time T and the lifetime of the second messenger τM are larger than the time that it takes for a cell to extend a pseudopod in the direction of the chemoattractant, which is 5–10 s in Dictyostelium (2729) and 10–15 s in neutrophils (31,32), i.e., equation M36 From this, it follows that equation M37

A second messenger that diffuses extremely slowly, such as transmembrane proteins (γ = 1, Dm = 0.03 μm2/s (11)), may have a very long lifetime τM up to 300 s before it diffuses outside the emerging pseudopod, and consequently the sampling fold I can be large (Fig. 5). However, diffusion is so slow that spatial integration of information over a distance of 1 μm would take several minutes and is essentially absent within the ~10 s time period that a cell extends a pseudopod toward a chemoattractant source. With a second messenger that diffuses somewhat faster, such as a phospholipid (γ = 1.5, Dm = 1 μm2/s; (33,34)), spatial integration of the activity of multiple receptors will occur over a distance of 1 μm in ~0.5 s, and over a distance of the emerging pseudopod in ~4 s. In addition, the sampling fold I can be >1, provided that the receptor dissociates fast with a lifetime of <~3s. With a second messenger that diffuses still faster, such as soluble compounds like cAMP, cGMP, or IP3 (γ = 2, Dm = 300 μm2/s (3537)), it is essentially impossible to retain spatial information, unless the lifetimes of the second messenger τM and the activated receptor τR are extremely short, in the order of 30 ms.

FIGURE 5
Dispersion length of diffusing second messengers as function of their lifetime and diffusion rate constant. The dispersion length of several potential second messengers was calculated with the equation equation M38where DM is the diffusion rate constants (D, in ...

In conclusion, the optimal second messenger for chemotaxis in eukaryotes has a lifetime of 3–20 s and diffuses with a rate constant of ~1 μm2/s. To obtain a sampling fold above 1, the chemoattractant should be detected by a receptor that has a lifetime of ~1 s. These properties allow for maximal temporal and spatial integration of activated receptors and minimal noise of stochastic receptor activation.

In Dictyostelium, the average lifetime of the cAMP-receptor complex τR is ~0.7–3.5 seconds (25,26); similar data have been obtained for the interaction between the fMLP and its receptor on human neutrophils, showing fMLP-receptor lifetimes between 0.5 and 3 s (38,39). Thus, the observed parameters for Dictyostelium are I = 1.8 (range 1.2–2.4) and τR = 0.7–3.5 s, which yield a hypothetical second messenger with a lifetime of τM = 1.5–7 s and a diffusion rate constant of DM = 0.25–4 μm2/s. The lifetime and diffusion rate constants of some potential second messengers have been measured. Although soluble second messengers such as cGMP and IP3 have the appropriate lifetime (40,41), they diffuse too fast (3537). In contrast, membrane-bound second messengers have generally slow diffusion (33,34) and some of them also short lifetimes (42). The best candidate is PIP3, which perfectly fits in the biophysical profile: very low basal levels resulting in low nonreceptor noise, rapid transient accumulation after cAMP stimulation with a half-life of ~5 s, and a diffusion rate constant of ~0.5 μm2/s, resulting in a dispersion length of ~2.5 μm (7,33,34,42). It should be mentioned, however, that inhibition of PI3kinase has only moderate effects on chemotaxis in Dictyostelium (29,4345), suggesting that other second messengers must be present that can take over the function of PIP3 in PI3kinase-null cells. Alternatives are the lipid products of PLC, PLA2, and PLD, or combinations of these enzymes, such as diacylglycerol, fatty acids (e.g., arachidonic acid) and lysophospholipids (e.g., lysophosphatidic acid).

APPENDIX

Equations A1–A6

When a pipette filled with cAMP is inserted in a field of Dictyostelium cells, cAMP will diffuse continuously from the pipette, leading a stable spatial gradient of which the concentration C(x) and the spatial gradient [nabla]C(x) are dependent on the distance (x) from the pipette according to

equation M39
(A1)

and

equation M40
(A2)

where Cp is the cAMP concentration in the pipette and α is a proportionality constant that depends on the geometry of the pipette and the applied pressure; the experimentally observed value in our experiments is α = 0.05 (see Materials and Methods). These equations are accurate descriptions of the cAMP gradient at a distance beyond 15 μm from the pipette; at shorter distances, more complex equations are required (M. Postma and P. J. M. van Haastert, unpublished).

In the modified Zigmond chamber, cells are deposited under a bridge. A cAMP gradient is formed by placing agar blocks with buffer on one side of the bridge and an agar block with cAMP on the other side (20). After a while, cells experience a gradient with the following properties (20):

equation M41
(A3)

and

equation M42
(A4)

where Cs is the cAMP concentration at the source, x is the distance from the source (700 μm), and L is the width of the bridge (2000 μm).

The small population assay scores the chemotactic response of a population of cells toward a cAMP source that has been applied at some distance away as a single dose. Due to diffusion in the agar, a transient cAMP gradient is formed that is maximal with the following properties (8):

equation M43
(A5)

and

equation M44
(A6)

where r is the radius of the cAMP droplet (150 μm), x is the distance between cell population and cAMP source (250 or 650 μm), and Cd is the applied cAMP concentration in the droplet. Note in these equations the different order in x compared to the pipette assay, which is due to the single dose of cAMP in the small population assay compared to continuous release of cAMP from the pipette.

Equation A7

The difference in number of occupied receptors in the front half of the cell directed toward the gradient and the back half of the cell is given by

equation M45
(A7a)

We are interested in chemotaxis at low cAMP concentrations (C [double less-than sign] KD) in shallow gradients (ΔC [double less-than sign] C) for which it holds that equation M46 and equation M47 are approximately equal to equation M48 Therefore Eq. A7a reduces to Eq. A7:

equation M49
(A7)

Equations A8 and A9

We assume that cAMP-binding to the receptor is a simple bimolecular reaction with association rate k1C and dissociation rate k1. The autocorrelation of the total receptor occupancy for lag time τ is given (46) by

equation M50
(A8a)

where Rt is the total number of receptors, equation M51 denotes the fraction of occupied receptors, equation M52 denotes the fraction of unoccupied receptors, and equation M53 denotes the macroscopic time constant for the binding reaction. At zero lag time τ = 0, the value of the autocorrelation function equals the variance of total receptor occupancy equation M54

When the signal of occupied receptors is averaged over time, the filtered signal RF(t) is the convolution with an averaging filter U(t) and the receptor occupancy R*(t):

equation M55
(A8b)

where U(t) is the rectangular filter with sampling time T:

equation M56
(A8c)

The autocorrelation function of the filter U(t) is

equation M57
(A8d)

The autocorrelation function of the filtered signal is then given by the convolution of the autocorrelation functions ARR and AUU:

equation M58
(A8e)

The variance of receptor occupancy equation M59during a sampling time T is then given by the variance of the filtered signal:

equation M60
(A8f)

which with equation M61 yields

equation M62
(A8g)

where

equation M63
(A9)

Thus, the variance of total receptor occupancy after sampling equation M64is the variance without sampling equation M65divided by I, the sampling fold.

In our experiments, we are interested in the noise of the system in shallow gradients at low cAMP concentrations (C [double less-than sign] KD), which implies that pu = 1 and thus Rtpopu = R*, and τR = 1/k−1, the lifetime of the cAMP-receptor complex. Therefore at low cAMP concentrations

equation M66
(2)

Notes

Editor: Jennifer Linderman.

References

1. Baggiolini, M. 1998. Chemokines and leukocyte traffic. Nature. 392:565–568. [PubMed]
2. Campbell, J. J., and E. C. Butcher. 2000. Chemokines in tissue-specific and microenvironment-specific lymphocyte homing. Curr. Opin. Immunol. 12:336–341. [PubMed]
3. Crone, S. A., and K. F. Lee. 2002. The bound leading the bound: target-derived receptors act as guidance cues. Neuron. 36:333–335. [PubMed]
4. Szurmant, H., and G. W. Ordal. 2004. Diversity in chemotaxis mechanisms among the bacteria and archaea. Microbiol. Mol. Biol. Rev. 68:301–319. [PMC free article] [PubMed]
5. Wadhams, G. H., and J. P. Armitage. 2004. Making sense of it all: bacterial chemotaxis. Nat. Rev. Mol. Cell Biol. 5:1024–1037. [PubMed]
6. Devreotes, P., and C. Janetopoulos. 2003. Eukaryotic chemotaxis: distinctions between directional sensing and polarization. J. Biol. Chem. 278:20445–20448. [PubMed]
7. Postma, M., L. Bosgraaf, H. M. Loovers, and P. J. M. van Haastert. 2004. Chemotaxis: signalling modules join hands at front and tail. EMBO Rep. 5:35–40. [PMC free article] [PubMed]
8. Mato, J. M., A. Losada, V. Nanjundiah, and T. M. Konijn. 1975. Signal input for a chemotactic response in the cellular slime mold Dictyostelium discoideum. Proc. Natl. Acad. Sci. USA. 72:4991–4993. [PMC free article] [PubMed]
9. van Haastert, P. J. M. 1983. Sensory adaptation of Dictyostelium discoideum cells to chemotactic signals. J. Cell Biol. 96:1559–1565. [PMC free article] [PubMed]
10. Xiao, Z., N. Zhang, D. B. Murphy, and P. N. Devreotes. 1997. Dynamic distribution of chemoattractant receptors in living cells during chemotaxis and persistent stimulation. J. Cell Biol. 139:365–374. [PMC free article] [PubMed]
11. Ueda, M., Y. Sako, T. Tanaka, P. Devreotes, and T. Yanagida. 2001. Single-molecule analysis of chemotactic signaling in Dictyostelium cells. Science. 294:864–867. [PubMed]
12. Berg, H. C., and E. M. Purcell. 1977. Physics of chemoreception. Biophys. J. 20:193–219. [PMC free article] [PubMed]
13. Bialek, W., and S. Setayeshgar. 2005. Physical limits to biochemical signaling. Proc. Natl. Acad. Sci. USA. 102:10040–10045. [PMC free article] [PubMed]
14. Arrieumerlou, C., and T. Meyer. 2005. A local coupling model and compass parameter for eukaryotic chemotaxis. Dev. Cell. 8:215–227. [PubMed]
15. Moghe, P. V., and R. T. Tranquillo. 1994. Stochastic model of chemoattractant receptor dynamics in leukocyte chemosensory movement. Bull. Math. Biol. 56:1041–1093. [PubMed]
16. Tranquillo, R. T., D. A. Lauffenburger, and S. H. Zigmond. 1988. A stochastic model for leukocyte random motility and chemotaxis based on receptor binding fluctuations. J. Cell Biol. 106:303–309. [PMC free article] [PubMed]
17. Fisher, P. R., R. Merkl, and G. Gerisch. 1989. Quantitative analysis of cell motility and chemotaxis in Dictyostelium discoideum by using an image processing system and a novel chemotaxis chamber providing stationary chemical gradients. J. Cell Biol. 108:973–984. [PMC free article] [PubMed]
18. Fisher, P. R. 1990. Pseudopodium activation and inhibition signals in chemotaxis by Dictyostelium discoideum amoebae. Semin. Cell Biol. 1:87–97. [PubMed]
19. Soll, D. R., D. Wessels, P. J. Heid, and H. Zhang. 2002. A contextual framework for characterizing motility and chemotaxis mutants in Dictyostelium discoideum. J. Muscle Res. Cell Motil. 23:659–672. [PubMed]
20. Veltman, D. M., and P. J. van Haastert. 2006. Guanylyl cyclase protein and cGMP product independently control front and back of chemotaxing Dictyostelium cells. Mol. Biol. Cell. 17:3921–3929. [PMC free article] [PubMed]
21. Konijn, T. M. 1970. Microbiological assay for cyclic 3′,5′-AMP. Experientia. 26:367–369. [PubMed]
22. Soll, D. R. 1999. Computer-assisted three-dimensional reconstruction and motion analysis of living, crawling cells. Comput. Med. Imaging Graph. 23:3–14. [PubMed]
23. Andrew, N., and R. H. Insall. 2007. Chemotaxis in shallow gradients is mediated independently of PtdIns 3-kinase by biased choices between random protrusions. Nat. Cell Biol. 9:193–200. [PubMed]
24. Song, L., S. M. Nadkarni, H. U. Bodeker, C. Beta, A. Bae, C. Franck, W. J. Rappel, W. F. Loomis, and E. Bodenschatz. 2006. Dictyostelium discoideum chemotaxis: threshold for directed motion. Eur. J. Cell Biol. 85:981–989. [PubMed]
25. van Haastert, P. J. M., and R. J. De Wit. 1984. Demonstration of receptor heterogeneity and affinity modulation by nonequilibrium binding experiments. The cell surface cAMP receptor of Dictyostelium discoideum. J. Biol. Chem. 259:13321–13328. [PubMed]
26. van Haastert, P. J. M., R. J. de Wit, P. M. W. Janssens, F. Kesbeke, and J. DeGoede. 1986. G-protein-mediated interconversions of cell-surface cAMP receptors and their involvement in excitation and desensitization of guanylate cyclase in Dictyostelium discoideum. J. Biol. Chem. 261:6904–6911. [PubMed]
27. Gerisch, G., D. Hulser, D. Malchow, and U. Wick. 1975. Cell communication by periodic cyclic-AMP pulses. Phil. Trans. R. Soc. London. B. 272:181–192. [PubMed]
28. Swanson, J. A., and D. L. Taylor. 1982. Local and spatially coordinated movements in Dictyostelium discoideum amoebae during chemotaxis. Cell. 28:225–232. [PubMed]
29. Postma, M., J. Roelofs, J. Goedhart, H. M. Loovers, A. J. W. G. Visser, and P. J. M. van Haastert. 2004. Sensitization of Dictyostelium chemotaxis by PI3-kinase mediated self-organizing signalling patches. J. Cell Sci. 117:2925–2935. [PubMed]
30. Postma, M., and P. J. M. van Haastert. 2001. A diffusion-translocation model for gradient sensing by chemotactic cells. Biophys. J. 81:1314–1323. [PMC free article] [PubMed]
31. Shelve, D. V., A. M. Alterative, and D. Chodniewicz. 2004. Controlled pseudopod extension of human neutrophils stimulated with different chemoattractants. Biophys. J. 87:688–695. [PMC free article] [PubMed]
32. Gerisch, G., and H. U. Keller. 1981. Chemotactic reorientation of granulocytes stimulated with micropipettes containing fMet-Leu-Phe. J. Cell Sci. 52:1–10. [PubMed]
33. Almeida, P. F. F., and W. L. C. Vaz. 1995. Lateral Diffusion in Membranes. Elsevier, Amsterdam.
34. Haugh, J. M., F. Codazzi, M. Teruel, and T. Meyer. 2000. Spatial sensing in fibroblasts mediated by 3′ phosphoinositides. J. Cell Biol. 151:1269–1280. [PMC free article] [PubMed]
35. Chen, C., T. Nakamura, and Y. Koutalos. 1999. Cyclic AMP diffusion coefficient in frog olfactory cilia. Biophys. J. 76:2861–2867. [PMC free article] [PubMed]
36. Tsaneva-Atanasova, K., D. I. Yule, and J. Sneyd. 2005. Calcium oscillations in a triplet of pancreatic acinar cells. Biophys. J. 88:1535–1551. [PMC free article] [PubMed]
37. Hofer, T., L. Venance, and C. Giaume. 2002. Control and plasticity of intercellular calcium waves in astrocytes: a modeling approach. J. Neurosci. 22:4850–4859. [PubMed]
38. Hoffman, J. F., J. J. Linderman, and G. M. Omann. 1996. Receptor up-regulation, internalization, and interconverting receptor states. Critical components of a quantitative description of N-formyl peptide-receptor dynamics in the neutrophil. J. Biol. Chem. 271:18394–18404. [PubMed]
39. Sklar, L. A., H. Mueller, G. Omann, and Z. Oades. 1989. Three states for the formyl peptide receptor on intact cells. J. Biol. Chem. 264:8483–8486. [PubMed]
40. van Haastert, P. J. M., and P. R. van der Heijden. 1983. Excitation adaptation and deadaptation of the cAMP mediated cGMP response in Dictyostelium discoideum. J. Cell Biol. 96:347–353. [PMC free article] [PubMed]
41. van Haastert, P. J. M., M. J. de Vries, L. C. Penning, F. Roovers, J. van der Kaay, C. Erneux, and M. M. van Lookeren Campagne. 1989. Chemoattractant and guanosine 5′-[gamma-thio]triphosphate induce the accumulation of inositol 1,4,5,-trisphophate in Dictyostelium cells that are labelled with [3H] inositol by electroporation. Biochem. J. 258:577–586. [PMC free article] [PubMed]
42. Huang, Y. E., M. Iijima, C. A. Parent, S. Funamoto, R. A. Firtel, and P. N. Devreotes. 2003. Receptor mediated regulation of PI3Ks confines PI(3,4,5)P3 to the leading edge of chemotaxing cells. Mol. Biol. Cell. 14:1913–1922. [PMC free article] [PubMed]
43. Funamoto, S., R. Meili, S. Lee, L. Parry, and R. A. Firtel. 2002. Spatial and temporal regulation of 3-phosphoinositides by PI 3-kinase and PTEN mediates chemotaxis. Cell. 109:611–623. [PubMed]
44. Iijima, M., and P. Devreotes. 2002. Tumor suppressor PTEN mediates sensing of chemoattractant gradients. Cell. 109:599–610. [PubMed]
45. Loovers, H., M. Postma, I. Keizer-Gunnink, Y. E. Huang, P. N. Devreotes, and P. J. M. Van Haastert. 2006. Distinct roles of PI(3,4,5)P3 during chemoattractant signaling in Dictyostelium: a quantitative in vivo analysis by inhibition of PI3-kinase. Mol. Biol. Cell. 17:1503–1513. [PMC free article] [PubMed]
46. Colquhoun, D., and A. G. Hawkes. 1977. Relaxation and fluctuations of membrane currents that flow through drug-operated channels. Proc. R. Soc. Lond. B Biol. Sci. 199:231–262. [PubMed]

Articles from Biophysical Journal are provided here courtesy of The Biophysical Society
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...