Cellular memory in eukaryotic chemotaxis depends on the background chemoattractant concentration

Richa Karmakar, Man-Ho Tang, Haicen Yue, Daniel Lombardo, Aravind Karanam, Brian A. Camley, Alex Groisman, and Wouter-Jan Rappel

Phys. Rev. E 103, 012402 – Published 6 January 2021

Abstract

Cells of the social amoeba Dictyostelium discoideum migrate to a source of periodic traveling waves of chemoattractant as part of a self-organized aggregation process. An important part of this process is cellular memory, which enables cells to respond to the front of the wave and ignore the downward gradient in the back of the wave. During this aggregation, the background concentration of the chemoattractant gradually rises. In our microfluidic experiments, we exogenously applied periodic waves of chemoattractant with various background levels. We find that increasing background does not make detection of the wave more difficult, as would be naively expected. Instead, we see that the chemotactic efficiency significantly increases for intermediate values of the background concentration but decreases to almost zero for large values in a switch-like manner. These results are consistent with a computational model that contains a bistable memory module, along with a nonadaptive component. Within this model, an intermediate background level helps preserve directed migration by keeping the memory activated, but when the background level is higher, the directional stimulus from the wave is no longer sufficient to activate the bistable memory, suppressing directed migration. These results suggest that raising levels of chemoattractant background may facilitate the self-organized aggregation in Dictyostelium colonies.

Received 21 June 2020
Accepted 16 December 2020

DOI:https://doi.org/10.1103/PhysRevE.103.012402

Physics Subject Headings (PhySH)

Cell aggregation Cell migration Cell signaling Chemotaxis

Physics of Living Systems

Authors & Affiliations

Richa Karmakar¹, Man-Ho Tang ¹, Haicen Yue ², Daniel Lombardo ¹, Aravind Karanam¹, Brian A. Camley ³, Alex Groisman¹, and Wouter-Jan Rappel ^1,*

¹Department of Physics, University of California, San Diego, La Jolla, California 92093, USA
²Courant Institute for Mathematical Sciences, New York University, New York, New York 10012, USA
³Department of Physics & Astronomy, Department of Biophysics, Johns Hopkins University, Baltimore, Maryland 21218, USA

^*rappel@physics.ucsd.edu

Article Text (Subscription Required)

Click to Expand

References (Subscription Required)

Click to Expand

Issue

Vol. 103, Iss. 1 — January 2021

Reuse & Permissions

Access Options

Article Available via CHORUS

Download Accepted Manuscript

Author publication services for translation and copyediting assistance advertisement

Images

Figure 1
(a) Left: Schematic of the microfluidic wave device, with observation region indicated by black box (scale bar: 3 mm). Right: Snapshot of cells moving on the micropatterned substrate, with symbols corresponding to cells identified by the machine learning algorithm (red circles: cells used in our analysis; blue X's: excluded cells that are too close to one another; the blurry spots are out-of-focus dirt particles and other irregularities that are not identified as cells by the machine learning algorithm; scale bar: $100 μ m$ ). (b)–(c) Spatial (b) and temporal profile (c) of the cAMP wave, determined from the fluorescent intensity of the dye, and the result of the Gaussian fit. (d)–(e) Images of the two substrate patterns used in this study, with green highlighting the location of the PEG-gel stripes. The pattern consists of either 4 narrow ( $\sim 10 μ m$ ) and 1 wide ( $\sim 25 μ m$ ) untreated stripes (d) or of sixth variable-width stripes, ranging from $\sim 6 μ m$ to $\sim 25 μ m$ (e). In both patterns, the untreated stripes are separated by $30 μ m$ wide nonadhesive PEG-gel stripes. Scale bar: $50 μ m$ . (f)–(g) The CI as a function of time for the current experiment using a micropatterned substrate (f) and in a previous study [23], using a nonpatterned substrate (g). The results are qualitatively similar, indicating that restricting the cells to 1D stripes does not affect their chemotactic behavior.
Reuse & Permissions
Figure 2
(a)–(i) Experimentally determined average CI as a function of time (measured relative to peak of wave) for different concentrations of background cAMP (0–150 nM). In each panel, gray dots represent the CI of individual cells, the black curve is the binned average over $N = 3 - 4$ different experiments, and the dashed red line is the cAMP concentration of the wave. (j) Average $x$ -component of the velocity of cells for different concentrations of background cAMP. Time is binned in intervals of 0.5 min. Error bars represent the standard error of the mean obtained using bootstrapping.
Reuse & Permissions
Figure 3
(a) Schematic diagram of the chemotactic model, consisting of a receptor $R$ , an activator $E$ , an inhibitor $I$ , a response element $S$ , and a memory component $M$ . Simulations are carried out in a 1D geometry (top drawing). As indicated by the bottom bar, $M$ is bistable, with a low and a high state, determined by parameters a and b. (b)–(e) Model results for different background cAMP concentrations added to a periodic wave, shown as a dashed red line. The black line represents the CI, the blue (light gray)/red (dash-dotted) line is the response $S$ at the front/back of the cell, and the dotted magenta line corresponds to the memory $M$ at the front.
Reuse & Permissions
Figure 4
(a)–(b) Comparison between experimental results from Ref. [35] (symbols) and model results. Shown are the maximum change in intensity $I_{peak}$ (a) and its corresponding time $T_{peak}$ (b) as a function of the uniform change in cAMP concentration for two different cAMP pretreatment concentrations. (c) Average CI (red symbols) versus background cAMP concentration.
Reuse & Permissions
Figure 5
(a) Steady-state value of $S_{f}$ as a function of uniform ${[cAMP]}_{bg}$ for the case $M_{f} = 1$ . (b) $S_{f}$ and $M_{f}$ as a function of time in the full model for a value of ${[cAMP]}_{bg}$ that does ( ${[cAMP]}_{bg} = 0.5$ nM; dashed lines) and does not ( ${[cAMP]}_{bg} =$ 0.4 nM; solid lines) result in persistent memory. (c) Maximum value of $S_{f}$ and $M_{f}$ in a wave as a function of ${[cAMP]}_{bg}$ . There is a sharp, switch-like transition at ${[cAMP]}_{bg} \approx 69$ nM. (d) Response of the full model for ${[cAMP]}_{bg}$ just below ( ${[cAMP]}_{bg} = 69$ nM; solid lines) and just above ( ${[cAMP]}_{bg} = 70$ nM; dashed lines) the switch-like transition.
Reuse & Permissions
Figure 6
Phase diagram in the $a − b$ space quantifying $〈 M_{f} - M_{b} 〉$ for three different values of ${[cAMP]}_{bg}$ . Three regions, with sharp transitions, can be identified: $〈 M_{f} - M_{b} 〉 = 1$ (yellow regions) for which the front memory is always high while the back memory is always low, $〈 M_{f} - M_{b} 〉 = 0$ (dark blue regions) corresponding to low memory at the front and the back, and intermediate values of $〈 M_{f} - M_{b} 〉$ for which the front memory is high during part of the wave. The values of $a$ and $b$ corresponding to our study are marked by a red X.
Reuse & Permissions

Physical Review E

covering statistical, nonlinear, biological, and soft matter physics