Are Single Cells Capable of a Simple Form of Learning?

Researchers at the Centre for Genomic Regulation (CRG), and at Harvard Medical School have used mathematical modeling to show how individual cells appear capable of learning, a behavior once deemed exclusive to animals with brains and complex nervous systems. Results from their studies, which focused on a simple form of learning known as habituation, could represent an important shift in how we view the fundamental units of life.

The findings offer evidence that even single-cell creatures, such as ciliates and amoebae, as well as the cells in our own bodies, could exhibit habituation akin to that seen in more complex organisms with brains, suggesting that single cells are capable of behaviors more complex than currently appreciated. If single cells can “remember,” then the results could also help explain how cancer cells develop resistance to chemotherapy or how bacteria become resistant to antibiotics—situations where cells seem to learn from their environment.

“Rather than following pre-programmed genetic instructions, cells are elevated to entities equipped with a very basic form of decision making based on learning from their environments,” said Jeremy Gunawardena, PhD, associate professor of systems biology at Harvard Medical School. “This finding opens up an exciting new mystery for us: How do cells without brains manage something so complex?”

Gunawardena is co-author of the team’s published study in Current Biology, titled “Biochemically plausible models of habituation for single-cell learning.” In their report the authors concluded, “Our results suggest that individual cells may exhibit habituation behavior as rich as that which has been codified in multi-cellular animals with central nervous systems and that the relative simplicity of the biomolecular level may enhance our understanding of the mechanisms of learning.”

The capacity to learn and to adapt is central to evolution and to survival. “The ability to learn is typically attributed to animals with brains,” the authors noted. Habituation— adaptation’s less-glamorous sibling—is the process by which an organism gradually stops responding to a repeated stimulus. It is why humans stop noticing the ticking of a clock, or become less distracted by flashing lights.

Up until recently, habituation was also deemed the exclusive domain of complex organisms with brains and nervous systems, such as worms, insects, birds, and mammals. “Habituation may be rationalized as a fundamental filtering mechanism, or regulator of attention, in systems exposed to multiple stimuli,” the team explained. “It is ‘‘non-associative’’ in requiring only a single stimulus, which elicits, upon repetitive presentation, a steadily declining response that reaches a plateau.” This change in response to the same stimulus is sometimes offered as “an informal, lowest common denominator definition of learning,” they suggested.

This lowest form of learning has been studied extensively in animals with complex nervous systems. “Habituation has been codified from studies in both invertebrate and vertebrate animals as having 10 characteristic hallmarks, seven of which involve a single stimulus,” the investigators continued. But whether learning-like behaviors like habituation exist at cellular scale is a question that’s remained fraught with controversy. Early 20th-century experiments with the single-celled ciliate Stentor roeselii first shed light on behavior that resembled learning, but the studies were overlooked and dismissed at the time. In the 1970s and 1980s, signs of habituation were found in other ciliates, and modern experiments have continued to add further weight to the theory.

Moreover, the authors pointed out, with reference to prior studies, “Because a single-cell organism must solve the same survival problems as any organism, in a world of  ‘‘blooming, buzzing confusion,’’ it may seem reasonable that evolution provided it with elementary forms of learning that are similar to those used by animals. Indeed, from an evolutionary perspective, we may even speculate that it was the former that gave rise to the latter.”

Co-author Rosa Martinez, PhD, a researcher at the Centre for Genomic Regulation (CRG) in Barcelona, said, “These creatures are so different from animals with brains. To learn would mean they use internal molecular networks that somehow perform functions similar to those carried out by networks of neurons in brains. Nobody knows how they are able to do this, so we thought it is a question that needed to be explored.”

Cells rely on biochemical reactions as their means of processing information. For example, the addition or removal of a phosphate tag from the surface of a protein causes it to switch on or off. To track how cells process information, instead of studying cells in a lab dish, the scientists used advanced computer simulations based on mathematical equations to monitor these reactions and decode the “language” of the cell. This allowed them to see how the molecular interactions inside cells changed when exposed to the same stimulus over and over again.

Specifically, the study looked at two common molecular circuits—negative feedback (NF) loops and incoherent feedforward (IFF) loops. In negative feedback, the output of a process inhibits its own production, akin to a thermostat shutting off a heater when a room reaches a certain temperature. In incoherent feedforward loops, a signal simultaneously activates both a process and its inhibitor, such as a motion-activated light with a timer. After detecting movement, the light automatically switches off after a certain period of time.

“NF and IFF are ubiquitous cellular motifs with distinctive properties,” the scientists stated.  “Unlike most other motifs, they characteristically show adaptation: a sustained stimulus elicits a response peak that then decays to a lower steady state.” The requirement for some form of response downregulation is common to both adaptation and habituation, the team continued, “… which suggested that the NF and IFF motifs were good starting points for studying habituation.”

The simulations suggested that cells use a combination of at least two molecular circuits to finetune their response to a stimulus and reproduce all the hallmark features of habituation seen in more complex forms of life. “Here, we show by mathematical modeling that simple molecular networks, based on plausible biochemistry with common motifs of negative feedback and incoherent feedforward, can robustly exhibit all single-stimulus hallmarks,” they said. One of the key findings is a requirement for “timescale separation” in the behavior of the molecular circuits, where some reactions happen much faster than others. “We think this could be a type of ‘memory’ at the cellular level, enabling cells to both react immediately and influence a future response” explained Martinez.

In this video, a single-cell pond dweller called Stentor roeselii exhibits markers of avoidance behavior, as reported in earlier research led by Gunawardena. The new study suggests this organism is also capable of habituation.

The finding may also illuminate a longstanding debate between neuroscientists and cognitive researchers. For years, these two groups have had different takes on how habituation strength relates to the frequency or intensity of stimulation. Neuroscientists focus on observable behavior, noting that organisms show stronger habituation with more frequent or less intense stimuli. Cognitive scientists, however, insist on testing for the existence of internal changes and memory formation after habituation has taken place. When following their methodology, habituation seems stronger for less frequent or more intense stimuli.

The study shows that the behavior of the models aligns with both views. During habituation, the response decreases more with more frequent or less intense stimuli, but after habituation, the response to a common stimulus is also stronger in these cases.  “The models reveal how the hallmarks arise from underlying properties of timescale separation and reversal behavior of memory variables, and they reconcile opposing views of frequency and intensity sensitivity expressed within the neuroscience and cognitive science traditions,” the team stated.

“Neuroscientists and cognitive scientists have been studying processes which are basically two sides of the same coin,” says Gunawardena. “We believe that single cells could emerge as a powerful tool to study the fundamentals of learning.”

The results add to a small but growing body of work on this subject. Earlier work led by Gunawardena found that a single-cell ciliate showed avoidance behavior, not unlike the actions observed in animals that encounter unpleasant stimuli.

The team’s newly reported findings do need to be confirmed with real-world biological data. Their study used mathematical modeling to explore the concept of learning in cells because it let them test many different scenarios rapidly to see which ones are worth investigating further in real experiments.

The work could lay the foundation for experimental scientists to now design lab experiments and test these predictions. “The moonshot in computational biology is to make life as programmable as a computer, but lab experiments can be costly and time-consuming,” said Martinez, who is based at the Barcelona Collaboratorium, a joint initiative between the CRG and EMBL Barcelona specifically designed to advance research based on mathematical modelling to address big questions in biology.

“Our approach can help us prioritize which experiments are most likely to yield valuable results, saving time and resources and leading to new breakthroughs,” she added. “We think it can be useful to address many other fundamental questions.”

Yet one daring idea would be to apply the concept of habituation to the relationship between cancer and immunity. Tumors are notoriously good evaders of immune surveillance because they trick immune cells into viewing them as innocent bystanders. In other words, the immune cells responsible for recognizing cancer may get somehow habituated to the presence of a cancer cell—the immune cell gets used to the stimulus and no longer responds to it.

“It’s akin to delusion. If we knew how these false perceptions get encoded in immune cells, we may be able to re-engineer them so that immune cells begin to perceive their environments correctly, the tumor becomes visible as malign, and they get to work,” Gunawardena said. “It is a fantasy right now, but it is a direction I would love to explore down the road.”