Evidence of conditioned behavior in amoebae
Associative memory is the main type of learning by which complex organisms endowed with evolved nervous systems respond efficiently to certain environmental stimuli. It has been found in different multicellular species, from cephalopods to humans, but never in individual cells. Here we describe a motility pattern consistent with associative conditioned behavior in the microorganism Amoeba proteus. We use a controlled direct-current electric field as the conditioned stimulus, and a specific chemotactic peptide as the unconditioned stimulus. The amoebae are capable of linking two independent past events, generating persistent locomotion movements that can prevail for 44 min on average. We confirm a similar behavior in a related species, Metamoeba leningradensis. Thus, our results indicate that unicellular organisms can modify their behavior during migration by associative conditioning.
One of the most remarkable accomplishments in the field of neuroscience is the description of essential principles that define the basic forms of associative memory. This fundamental cognitive property occurs through complex biological mechanisms by which the connection between two previously unrelated stimuli, or a behavior and a stimulus, is learned; when such process takes place, it is assumed that the association of these stimuli is stored in a memory system1. For centuries, different thinkers have shaped a very plentiful and venerable history of research on basic learning processes. The combined work of philosophers, naturalists, physiologists, and life scientists has set the baseline upon which the modern learning theory currently stands2. The most basic type of associative learning is the classical conditioning developed by the Nobel Prize Laureate Ivan Pavlov, who established the first systematic study of the fundamental principles of associative memory. In his studies, after an appropriate conditioning, dogs deprived of food were able to exhibit a consequent response -salivation- when a bell rung3.
Associative conditioning is ubiquitous in complex organisms endowed with evolved nervous systems, including all major vertebrate taxa and several invertebrate species4. This complex process can also be reproduced and analyzed in artificial neural networks and different computational models5. Conditioned learning confers to the organisms the ability to adapt to ever-changing environments and is considered a milestone for life’s survival. Despite its importance, associative conditioning has never been observed in individual cells.
In order to determine whether associative conditioned responses are involved in systemic cellular behaviors, we analyzed the movement trajectories of Amoeba proteus under two external stimuli by using an appropriate electric field as the conditioned stimulus and a specific peptide as chemo-attractant.
Amoebae represent an immensely diverse family of eukaryotic cells that can be found in nearly all habitats and constitute the major part of all eukaryote lineages6. Concretely, Amoeba proteus is a large free-living predatory amoeba with a notable capacity to detect and respond to chemical and physical cues allowing it to locate and consume near prey organisms such as bacteria and other protists.
These cells are able to migrate on flat surfaces and in three-dimensional substrate by a process known as amoeboid movement, which consists in pseudopodia extensions, cytoplasmic streaming, and flowing into these extensions changing permanently the cellular shape7.
Amoeboid locomotion represents one of the most widespread forms of cell motility and constitutes the typical way of locomotion in broad range of adherent and suspended eukaryotic cell types7. In mammalian cells, amoeboid locomotion is vital for multiple physiological processes as the development of the embryo8, the action of the immune system9 and the repair of wounds10. Likewise, it is also responsible for the spread of malignant tumors11.
The large free-living amoeba, Amoeba proteus, has served as a classic unicellular organism in many investigations for more than one hundred years12,13, mainly as a cellular model to study cell motility, membrane and cytoskeleton function, and the role of the nucleus14–16. However, despite the many investigations carried out so far, numerous biological aspects of this organism still remain poorly studied. On the other hand, diverse experimental studies have shown that Amoeba proteus exhibit robust galvanotaxis17, a directed movement in response to an electric field; in fact, it has been described that practically 100% of the amoebae migrate towards the cathode for long periods of time under a strong direct-current electric field in a range between 300 mV/mm and 600 mV/mm. Likewise, amoebae are known to display chemotactic behaviors; in particular, the peptide nFMLP, typically secreted by bacteria, is able to provoke a strong chemotactic response in many different types of cells. The presence of this peptide in the environment may indicate to the amoeba that food organisms might be nearby18. Given the large number of investigations carried out on this organism, the robustness in their behavior, easy handling in the laboratory, the relatively fast rate of migration (cells move at ~300 μm/min19) and the well documented sensitivity to electric fields and chemoattractants, we have chosen Amoeba proteus as the experimental study species in our work.
Here, we describe the emergence of an associative conditioned behavior in Amoeba proteus which corresponds to a new type of systemic migration pattern in the cell. This conditioned migration behavior seems to be an evidence of a primitive type of associative memory in a unicellular organism. In a preliminary study, we also confirm a similar behavior in a related species, Metamoeba leningradensis.
All our experiments have been carried out on a specific set-up that allowed us to expose the amoebae to both stimuli -galvanotaxis and chemotaxis- simultaneously. This system consists of two standard electrophoresis blocks, about 17.5 cm long, one directly plugged into a normal power supply and a second one connected to the first one via two agar bridges that transfer the current from one block to the other while preventing the direct contact of both the anode and the cathode with the medium where the cells were located (see Fig. 1, Supplementary Data 1, and data available in Methods section). On the central platform of the second electrophoresis block, we placed the experimental chamber, a sliding glass structure that enabled the creation of a laminar flux which not only allowed the electric current to pass through, but also generated an nFMLP peptide gradient that the amoebae were able to detect and respond to. In addition, when the sliding glass structure was opened, the placement and collecting of the cells was possible. We confirmed the establishment of the nFMLP gradient by the direct measurement of fluorescein-tagged peptide concentration with a plate reader. As shown in Fig. 2, the concentration of peptide in the middle part of the glass chamber (where the amoebae are placed) increases immediately following the flow establishment (within 2 min the concentration rises from zero to approximately 0.2 μM) and this concentration increases further (to 0.6 μM) for at least 30 min.
In the experiments, the cells were placed in the middle of the glass set-up and their displacements were monitored in small groups (see Methods section), being the individual trajectories recorded during periods of 30 min by using a camera connected to a microscope. The migration of 615 A. proteus and 210 M. leningradensis was quantitatively analyzed (Supplementary Data 2). All the experiments were carried out in Chalkley’s medium, a standard, nutrient-free saline medium at ambient temperature.
First, we recorded the locomotion trajectories of 50 amoebae (experimental replicates: 7, number of cells per replicate: 5–11) without any external influences (Fig. 3a). Under this condition, cell migration can be described as a correlated random motion characterized by low intrinsic directionality and progressive decreasing over time of the initial direction of migration20. In Fig. 3a, a representative example of amoebae locomotion in the absence of stimuli is depicted; cells exhibited significant changes in the movement guidance, and after 30 min they had explored practically all the directions of the experimentation chamber. The directionality of each cell was quantified by the cosine of the displacement angle17 (see Methods section), and the quantitative results showed that the values ranged between −0.987 and 1, being −0.125/1.55 the media/interquartile range (IQ), which indicated that in the absence of stimulus, these cells moved randomly without any defined guidance. In addition, the analysis of the distribution of displacement angles (i.e., the angle formed between the origin and the end of the movement, measured in radians) also confirmed no preference towards a certain direction (Fig. 3d).
Next, the galvanotactic locomotion of 50 cellular trajectories (experimental replicates: 8, number of cells per replicate: 4–8) was analyzed under an external, controlled direct-current electric field of about 300 mV/mm (Methods section). Our experimental observations (Fig. 3b) indicated that practically all the amoebae migrated towards the cathode for 30 min. These results matched with other previously reported experiments17. Cellular locomotion under this electric condition was characterized by stochastic movements with robust directionality, and cells exhibited a locomotion pattern tending to move in the direction of the immediately preceding movement by conserving their polarity in time towards the cathode. Taking as reference the experiment of Fig. 3b, the quantitative analysis indicated that the values of the cosines of displacements were distributed between 0.037 and 0.999 (0.993/0.03 median/IQ) (Fig. 3e). This result verified that a unique fundamental behavior characterized by an unequivocal directionality towards the cathode had emerged in the experimental system. The significance of our analysis was validated with a non-parametric test (Wilcoxon rank-sum test) comparing the distributions of the cosines of the displacement in both situations, without stimulus and under the presence of the electric field. The test (p = 10−14; Z = 7.442, Wilcoxon rank-sum test) corroborated that the behaviors without and with the stimulus (the electric field) were significantly different.
Here, the behavior of 50 amoebae (experimental replicates: 10, number of cells per replicate: 4–6) was analyzed during biochemical guidance by exposing the cells for 30 min to an nFMLP peptide gradient placed in the left side of the set-up. The experiment showed that 86% of exposed cells migrated towards the chemotactic gradient (Fig. 3c). In other words, the chemical gradient in the environment provoked in most amoebae a systemic behavior characterized by stochastic locomotion movements with robust directionality towards the attractant stimulus (the peptide)21. The cosines of the displacement angles of individual trajectories ranged between −0.997 and 0.986 (−0.825/0.72 median/IQ) (Fig. 3f). This result indicated that a single fundamental behavior characterized by a movement towards the peptide prevailed in the cells. The comparison between the cosine values obtained with and without chemotactic stimulus (p = 10−4; Z = −3.878, Wilcoxon rank-sum test) on one hand, and between the cosine values with chemotactic gradient and with the presence of electric field (p = 10−17; Z = 8.428, Wilcoxon rank-sum test) on the other, corroborated that the systemic locomotion behavior under the chemotactic gradient was totally different to both, the absence of stimulus, and the presence of an electric field.
Once the migrations of the amoebae in the three previous basic and independent experimental conditions (without stimulus, under galvanotaxis and under chemotaxis) were analyzed, we studied the trajectories of 180 Amoeba proteus (experimental replicates: 32, number of cells per replicate: 4–10) when they were exposed simultaneously to galvanotactic and chemotactic stimuli for 30 min (Fig. 4a). For such a purpose, we arranged the cathode on the right of the set-up and the anode with the nFMLP peptide solution on the left. The analysis of the amoebae trajectories showed that 53% of the cells ignored the electric field signal and moved towards the anode-peptide (23% of them exhibited a very sharp directionality), while the remaining 45.33% migrated to the cathode. Three cells (1.67%) presented an atypical behavior, remaining immobile but adhered to the substrate during the 30 min of the test, and therefore were included in the unconditioned group. The cosines of the displacement angles were distributed between −1 and 1 (−0.26/1.8 median/IQ). This analysis verified that two fundamental cellular migratory behaviors had emerged in the experimental system, one towards the anode and another towards the cathode. The statistical analysis confirmed the presence of these different behaviors (p = 10−30; Z = −11.435, Wilcoxon rank-sum test). In Fig. 6, a galvanotactic control of the cells that responded to the cathode during the induction process is shown. All the amoebae again migrated towards the cathode, confirming that these cells were unconditioned.
To verify if the cells that moved towards the anode during the induction process (Fig. 4a) present some kind of persistence in their migratory behavior, we analyzed 160 amoebae in three different scenarios in which they were exposed to several types of perturbations.
In the first scenario, 85 amoebae (experimental replicates: 32, number of cells per replicate: 1–7) that had previously migrated towards the anode-peptide during the exposition to two simultaneous stimuli (induction process) were manually extracted and placed for 5 min on a normal culture medium (Chalkley´s medium) in a small Petri dish in absence of stimuli. Then the cells, were deposited on a new identical glass and block set-up that had never been in contact with the chemotactic peptide nFMLP and exposed for the second time to a single electric field, without peptide, during 30 min (note that the total time after the first induction process was 35 min). The analysis of the individual trajectories showed that 82% of the cells ran to the anode where the peptide was absent (Fig. 4b). The cosines of displacements ranged between −1 and 0.998 (−0.854/0.77 median/IQ). This result supported mathematically that the majority of cells moved towards the anode in the absence of peptide, thus corroborating that a new locomotion pattern had appeared in the cells. Such systemic behavior (migration towards the anode in the absence of peptide) had never appeared before. The comparison between the cosines of displacements obtained during the galvanotaxis without previous induction (Fig. 3b) and the galvanotaxis after the induction (Fig. 4b) showed that this newly acquired cellular behavior is extremely unlikely to be obtained by chance (p = 10−19; Z = 8.878, Wilcoxon rank-sum test). Four cells in this scenario exhibited eventual displacements towards both sides without any preference, and therefore were included in the unconditioned group. Since 43 cells persisted in the migration towards the anode until the end of the galvanotaxis, we used 50 cells for the next step.
In the second scenario, we studied 50 amoebae (experimental replicates: 27, number of cells per replicate: 1–4) previously exposed to the induction process and to the conditioning test of the first scenario (5 min without stimulus and 30 min of galvanotaxis). After that, these cells were placed once more in Chalkley´s medium without any stimulus for 3 min, and then they were again exposed to galvanotaxis for 30 min but in this case the polarity of the electric field was inverted (the cathode was positioned where the anode was previously and vice versa). In short: after the induction process, the cells were exposed to 5 min without stimulus, 30 min on galvanotaxis, 3 min without stimulus and 30 min on galvanotaxis with inverted polarity; in total, the time elapsed between the end of the induction process and the end of the study was 68 min.
When an Amoeba proteus is placed in an electric field for long periods of time, the probability of dying or, at least, detaching from the substrate and adopting a spherical shape increases sharply. Therefore, the cells were physically extracted and replated for 3 min to minimize cell damage. Next, we changed the medium in the set-up, after that the cells were again placed in the clean experimental chamber, and finally exposed to galvanotaxis with inverted polarity with fresh medium for 30 min. By inverting the electric field, we demonstrated that the amoebae were neither directed to a specific point in the space nor they associated a specific point in the space to the peptide.
Even under these new and strict conditions, 58% of the cells continued migrating towards the anode (now positioned in the opposite side) thus maintaining the conditioned behavior (Fig. 4c). The remaining cells (42%) lost this ability. The comparison between the cosines of displacements in the galvanotaxis without induction (Fig. 3b) and the cosines from the second scenario also indicated that it is unlikely to obtain this new conditioned systemic behavior by chance (p = 10−11; Z = −6.491, Wilcoxon rank-sum test). Four cells displayed an atypical behavior characterized by immobility and adhesion to the substrate during the 30 min of the test, and three more cells showed eventual displacements towards both sides; all of them were included in the unconditioned group. Since 16 cells maintained the migration towards the anode until the end of the galvanotaxis, we used 25 cells for the next step.
In the third scenario, 25 cells (experimental replicates: 9, number of cells per replicate: 1–7) were exposed to the induction process and to the conditioning test of the first scenario (5 min without stimulus and 30 on galvanotaxis). Next, once again, the cells were placed in Chalkley´s medium without any stimulus, in this occasion for about 30 min, and then exposed to galvanotaxis with inverted polarity for 30 min like in the second scenario (Fig. 4d). In short, 5 min without stimulus, 30 min on galvanotaxis, 30 min without stimulus and 30 min on galvanotaxis with inverted polarity; therefore, the total time elapsed between the end of the induction process and the end of the study was 95 min.
Like in the second scenario, cells were physically extracted and re-plated for 3 min to minimize cell damage due to the long standing in an electric field. Next, we changed the medium in the set-up, then the cells were again placed in the clean experimental chamber, and finally exposed to galvanotaxis with inverted polarity with fresh medium for 30 min.
Even under these stricter conditions, 56% of the cells still maintained the migration to the anode, positioned in the opposite side, making evident that the new systemic behavior exhibits a remarkable robustness despite the perturbations introduced in the experiment. In addition, the test comparing the cosines during the third scenario (Fig. 4d) and the galvanotaxis without previous induction (Fig. 3b) showed that it is completely unlikely to obtain this newly acquired cellular behavior by chance (p = 10−11; Z = 6.221, Wilcoxon rank-sum test). One cell presented an atypical behavior characterized by immobility, and therefore was included in the unconditioned group.
A very restrictive criterion was adopted in these three scenarios: a cell was considered to present lasting directionality towards the anode only if after 15 or more minutes, the amoeba was still migrating towards the anode (which corresponds to 20 min after the induction process since all the amoebae were placed in Chalkley’s medium without any stimulus for 5 min after the induction) or if after moving initially towards the cathode, the cell corrected its trajectory showing a clear lasting directionality towards the anode. Note that these behaviors were never observed in any of the five galvanotactic experiments (Figs. 3b, 5 and 6a–c).
The new emergent systemic behavior is also characterized by a limited duration through time. Figure 7a is an illustrative example of the loss of conditioning in 15 induced cells (experimental replicates: 4, number of cells per replicate: 3–6) as time goes on. To quantify this phenomenon, we have measured the duration time of the conditioned behavior in the 148 conditioned cells in the three scenarios previously described.
In the first scenario, (35 min after the induction process, 81 conditioned cells) the analysis of the persistence level showed that 11 cells lost the inducted behavior at the beginning of the test, 27 cells did it after 20–33 min, whereas the rest (43 cells, 53%) maintained migration towards the anode until the end of the experiment (see Fig. 7b for details).
In the second scenario, (68 min after the induction process, 43 conditioned cells) 14 cells lost the persistent behavior at the beginning of the test and 13 did it after 52–66 min, whereas the remaining 16 cells (37%) continued the migration towards the anode until the end of the experiment (Fig. 7b).
In the third scenario, (95 min after the induction process, 24 conditioned cells) 10 cells lost the persistent behavior at the beginning of the test, 6 did it after 47–90 min, and the remaining 8 cells (33%) continued the migration towards the anode until the end of the experiment (Fig. 7b).
Despite all the perturbations introduced in the experiments, the whole analysis indicates that the average time of the cells that lost the acquired motility pattern during any of the three scenarios was 44.04 ± 21.8 min.
In order to examine the robustness of the observed conditioned behavior we have performed a preliminary study on another unicellular species, Metamoeba leningradensis, under the same induction process as the Amoeba proteus. The metamoebae were exposed to the same intensity of the electric field and to the same peptide (nFMLP) concentration, and therefore, values of amperage or optimal concentration of peptide were not adapted to generate more efficient responses of these organisms to such stimuli.