Conquering fear

The research findings of Sumantra Chattarji and Supriya Ghosh may lead to a potential therapy for post-traumatic stress disorder.

Sumantra Chattarji.

PAVLOVIAN reflex is a term that one is familiar with, at least intuitively. It refers to a certain survival instinct that leads one to take evasive action in response to some sensory signal—auditory, visual, touch—because of its association with some adverse occurrence or traumatic event in the past even if it does not present any danger in the immediate present. This is called fear conditioning and is the result of learning by association. The ingrained memory of the past signals potential danger and the brain triggers an action of avoidance as the best strategy.

But, while memories do serve as survival tools, this survival instinct can sometimes over time transform into a condition of generalised fear of stimuli that are only remotely linked to the aversive signal of the past. This failure of the brain to discriminate between safe and dangerous signals leads to a psychological syndrome known as the post-traumatic stress disorder (PTSD), a debilitating disorder triggered by some highly traumatic incident in the past.

One of the main symptoms of PTSD is “excessive generalisation of fear” where the victim not only displays a stronger fear response to situations that signal danger but also displays a fear response even to safe situations. Recently, some neuroscientists proposed this impaired learning—because of which there is a loss of ability to discriminate between what is potentially dangerous and what is safe—as a biomarker for PTSD.

Clinical evidence has linked such behaviour to hyperactivity in a region of the brain called amygdala, which is located deep within the brain and is the seat of complex emotional responses—fear, anxiety and aggression—and the long-term memories linked to them. However, the underlying mechanism in the brain has not been properly understood as techniques of imaging the nerve cells (neurons) do not have the necessary resolution to delineate clearly the responses of individual neurons.

Unravelling the neuronal basis The recent work of Sumantra Chattarji, a professor of neurobiology at the National Centre for Biological Sciences (NCBS) of the Tata Institute of Fundamental Research (TIFR) in Bangalore, and his former student Supriya Ghosh, who is currently at the University of Chicago, United States, marks a major advance in that direction. They have succeeded in unravelling the neuronal basis and the biochemical pathway that results in such hyperactive behaviour in the amygdala triggered by some earlier traumatic episode. Their findings may even lead to potential therapy for PTSD.

Based on fear conditioning experiments done on rats, the work of Chattarji and Ghosh provides fresh insights into how the brain loses its ability to distinguish between safe and dangerous stimuli, resulting in a state of generalised fear. Their research paper was published on December 1 in the online version of the journal Nature Neuroscience .

“I can’t get the memories out of my mind.… I am right back in Vietnam, in the middle of the monsoon season at my guard post. My hands are freezing, yet sweat pours from my entire body…. I smell a damp sulphur smell. Suddenly I see what’s left of my buddy Troy, his head on a bamboo platter, sent back to our camp by the Viet Cong.”

These are the words of a Vietnam War veteran of the United States Army. Whenever this former soldier hears a clap of thunder or touches a bamboo mat or sees an oriental woman, he gets into this hyperactive state of intense fear triggered by intense flashbacks of the traumatic incident that happened decades ago. This, as the NCBS press release on the Chattarji-Ghosh work points out, is an example of generalised fear conditioning. In fact, PTSD was originally described as “shell shock” in soldiers. Such conditioning has subsequently been clinically observed in victims of sexual violence, accidents, natural disasters and terrorism, according to the release.

In his classic experiments on associative learning that the Russian physiologist Ivan Pavlov carried out in the early 20th century with dogs, the animals initially did not respond to the sound of a bell. However, when the same sound was followed by food several times, with associative learning the dogs now displayed a state of conditioned response to the sound of the bell, which was previously a neutral stimulus. The dogs began salivating at the mere sound of the bell—in the expectation of food—even when the sound was not accompanied by food.

Interestingly, Pavlov also found that the dogs salivated not only to the original sound but also to similar sounds, particularly to those tones with frequencies close to the original, which indicated the tendency of the animals to generalise to a wider range of stimuli because they expect the sounds to produce a similar consequence, namely food.

“If the consequence,” Chattarji and Ghosh point out in the press release, “is not a reward but a painful punishment or a dangerous situation, then the stakes are obviously higher. If an animal under-generalises it may overlook future signs of danger, whereas if it over-generalises it may be too afraid to explore and thereby miss opportunities for feeding, mating, etc. So, striking the right balance is essential for survival.”

In their study, Chattarji and Ghosh used a variant of Pavlov’s experiments. They exposed a dozen rats to two distinctly different sounds, with one of them paired with a mild electric shock in the feet. Thus, one tone signalled danger and the other safety. By interweaving 10 of the footshock-paired danger tone with 10 of the safe tone, the scientists trained the rats trained to distinguish between the two sounds. The rats were found to learn quickly (after 24 hours) to discriminate between the dangerous and safe sounds. They displayed a clear fear response (freezing of their body movements) against the shock-associated tone but not against the safe tone.

To study the neuronal basis of this behaviour, the scientists also recorded the electrical signals from the individual neurons in the amygdala. The conditioned response resulted in an altered electrical activity of the amygdala neurons, and the changes accurately reflected the animal’s behavioural responses to the different sounds. The firing in the individual neurons was stronger in response to the dangerous tone compared with the safe tone. There was a clear correlation between the brain’s internal activity and the actions of the animal as a whole.

The study revealed that Pavlovian fear conditioning involved three distinct classes of neuronal activity. About 43 per cent of the neurons were “cue-specific”; that is, they displayed selective firing in response only to the shock-linked tone. About 6 per cent were “generalised cells”, which failed to discriminate between the two tones and fired equally strongly to both the tones. The remaining 51 per cent of the cells were “non-conditioned” cells, which did not show any response to either of the tones. The average activity of the amygdala neurons was thus an increased selective response to the shock-linked tone.

Neuronal basis of fear To study the neuronal basis of generalised fear, Chattarji and Ghosh repeated the experiment with a stronger footshock that was twice as intense as that in the weak-conditioning experiment. Their approach was guided by theoretical models of fear conditioning that suggest that, rather than using tones similar to the dangerous tone as Pavlov did, fear conditioning also gets enhanced by increasing the intensity of the dangerous tone itself. The scientists chose a different set of 11 rats for this strong-conditioning experiment.

In a manner similar to the weak-conditioning experiment, when the rats’ behaviour was observed 24 hours after they had been conditioned with a stronger danger tone and the safe tone, the scientists found that the rats’ ability to tell the dangerous tone from the safe tone had significantly deteriorated. It was found that even the safe tone, which was never paired with a shock, elicited a significantly higher level of locomotor freezing. The animal was not as selective as it was with weak conditioning, an indication of a tendency of conditioned fear becoming heightened or more generalised.

Correspondingly, at the level of neurons, while here too the activity could be classified into three categories, there was a notable change in the pattern of firing. The relative proportions of neurons in the different categories were markedly different from the weak-conditioning experiment (see figure). For example, the scientists found that while the proportion of cue-specific cells had dropped from 43 per cent to 32 per cent, the proportion of generalised cells—which displayed equal response to both the tones—had increased from the earlier 6 per cent to 30 per cent. That is, neurons that were once capable of discriminating were less discriminatory now as the animal began to exhibit increasingly generalised fear.

With the strong fear conditioning, the average neuronal activity was no longer tilted towards cue-specific stimulus. In fact, the increase in firing caused by the safe tone was comparatively higher than that evoked by the threatening tone. The neuronal activity thus once again mirrored the generalised behavioural response (heightened freezing) seen in the animals.

To understand more precisely the process of transition towards generalised fear at the neuronal level in the amygdala, Chattarji and Ghosh observed the neuronal response pattern with the neurons of a single animal instead of studying the responses of pooled neurons of a group of rats. They compared the neuronal activity in the same animal before and after their behavioural inability to discriminate between dangerous and safe tones following strong fear conditioning. They carried out these animal-by-animal experiments on eight rats. While the results of these experiments were consistent with the results obtained from pooled neurons, they also provided insights into the exact nature of the change in the pattern of responses among the amygdala neurons.

The scientists detected three kinds of changes in the neuronal firing pattern towards increased generalisation corresponding to the changes at the behavioural level in the single animal. From the majority group of neurons that were cue-specific during weak conditioning, they found that a subset lost their specificity and began firing in response to the safe tone as well and thus became generalising neurons. The second transition among the neurons that contributed to this generalisation was a significant subset of non-conditioned cells—which showed no response during weak conditioning to either of the tones—becoming generalised; that is, they began responding to both the sounds. Interestingly, another subset of non-conditioned cells became cue-specific!

“These neurons lack the ability to respond to weak conditioning but when they learn with strong conditioning, they learn correctly and become cue-specific,” pointed out Chattarji. “We do not know if these cells are inherently different in some way that they behave in this fashion. Also, this transition is probably a way of compensating for the loss of some cue-specific cells that became generalised under strong conditioning,” he added. Thus, even at the level of a single animal, the experiment showed that the three kinds of shifts together tilted the balance towards an increased generalised response among the neurons following strong fear conditioning.

To ascertain that the generalisation that they were observing was a true effect of the strong fear conditioning, the scientists conditioned a separate group of rats twice with the same weak-conditioning footshock. But this was seen not to result in any generalised response both at the behavioural and neuronal level. This, the scientists say, rules out the possibility that simply reconditioning can increase the probability of generalised fear. The fact that strong conditioning leads to a more generalised fear response is perhaps indicative of an intense traumatic event resulting in a predisposition towards a state of generalised fear. “Of course, this may not be true in all cases; it would vary from individual to individual. But it could be a contributory factor,” said Chattarji.

The above findings clearly demonstrate that the potentially greater cost of failing to discriminate correctly pushed the animals’ behaviour towards playing it safe. They behaved as if there was potential danger in the safe sound as well. The neurons too reflected this tendency to play it safe at the behavioural level. The work by Chattarji and Ghosh thus marks a significant advance in our understanding of how information is processed in the amygdala at the level of individual neurons to maintain a balance between when one should be and should not be afraid.

Since the fear-conditioning stimulus in this case is an auditory signal, the scientists also investigated the possible role of the auditory cortex in modulating the transition towards generalised fear. In the auditory system, the processed auditory information first reaches the thalamus region of the brain before being relayed to the auditory cortex (the region in the cerebral cortex which receives the auditory information) through a couple of more synapses. Earlier experiments by others had suggested that both the thalamic and the cortical inputs to the amygdala can contribute to fear conditioning.

While reversible inactivation of the amygdala by directly infusing muscimol (a mushroom toxin that prevents neurons from firing) into the amygdala prevented generalisation of conditioned fear, a similar targeted infusion of muscimol into the auditory cortex had no effect. This demonstrated that neuronal activity in the amygdala, but not the auditory cortex, is necessary for the transition into the generalised state of conditioned fear.

However, their investigations showed that the thalamic inputs to the amygdala seem to have a role in fear generalisation. The thalamic inputs to the amygdala differ from those from the auditory cortex in essentially two ways. One, the thalamic inputs typically have shorter delay times between stimulation and response (< 20 milliseconds) compared with the latency times involved in the auditory cortical response. Two, the auditory cortical response is more nuanced as compared to the ability of the thalamus to resolve sounds.

What the scientists discovered was that during fear generalisation, the short latency auditory inputs to the amygdala from the thalamus became stronger. That this quicker response to the auditory stimulus should be involved in fear generalisation rather than the slower and more nuanced mechanism is yet another confirmation that animals tend to err on the side of caution for survival, points out Chattarji.

“They would rather choose the low road to fear, even if it is wrong, rather than the high road to fear which could prove to be costly. This also fits in with an evolutionary perspective on fear generalisation,” he added. This picture, according to Chattarji, could also be valid for other sensory stimuli, such as light or smell. However, doing experiments with rats using auditory stimuli is relatively easier and unambiguous compared with other sensory inputs because of which conditioning with auditory signals has been the dominant model of investigations, Chattarji pointed out. Chattarji and Ghosh also looked into the underlying biochemical pathways that would induce hyperactive firing in amygdala neurons leading to a state of generalised fear in animal behaviour. It is known that activation of what is known as the cyclic adenosine monophosphate (cAMP)-protein kinase A (PKA) pathway enhances firing of nerve cells and synaptic transmission of chemical signals inducing excitation. This signalling pathway can be stimulated by a chemical called forskolin. The scientists examined whether cAMP-PKA signalling induced by administering forskolin directly into the amygdala of the animals resulted in the transition to generalised fear.

The researchers found that weak conditioning, in conjunction with simultaneous in vivo activation of cAMP-PKA signalling in the amygdala led to increased fear generalisation comparable to what was seen in strong-conditioning experiments. The forskolin-induced activation of cAMP-PKA signalling not only amplified fear generalisation at the behavioural level but also enhanced the firing of neurons triggered by tone and shock.

The above finding could serve as the basis for therapeutic interventions in PTSD cases. “Possible therapy could be in the nature of chemical blockers that target the receptors involved in this signalling pathway,” said Chattarji.