Survival is the prime directive of every animal, and it takes work. Animals are under constant threat from starvation, thirst, extreme temperatures, and life-threatening injury and illness. But when multiple threat signals compete for attention, organisms must prioritize the most urgent among them and adjust behaviors accordingly.
Two recent papers tackle the encoding of pain, hunger, and other survival threats in the nervous system, with each study identifying neural signaling at the parabrachial nucleus (PBN). This is a key hindbrain structure where sensory signals are integrated with information from higher brain centers to assess threats and prioritize behaviors.
The first study, from Nicholas Betley and colleagues, University of Pennsylvania, Philadelphia, US, identifies a specific subpopulation of hypothalamic neurons that project to the PBN to dampen inflammatory—but not acute—pain during extreme hunger. The study was published March 22 in Cell.
Perry Fuchs, University of Texas at Arlington, US, who was not involved in the study, called it “a very important paper with broad implications for the field, beyond this circuit.”
The second study, from Richard Palmiter and colleagues, University of Washington, Seattle, US, identifies a population of neurons in the PBN that serve as a detector of multiple threats. The neurons responded to acute pain, itch, visceral pain, and even non-noxious stimuli that pose a potential threat, such as a novel food. The study appeared March 29 in Nature.
Reza Sharif-Naeini, who was not involved in either study, wrote in a comment for PRF, “The [Nature] study by [first author] Carlos Campos and colleagues follows a long series of outstanding research from the Palmiter group in their attempts to understand the physiological roles of subsets of neurons in the parabrachial nucleus.” (See full comment and accompanying figure from Sharif-Naeini below).
A threat hierarchy
Betley’s group has been studying neurons that control feeding behavior for years, but they wanted to study how hunger interacted with other stressors. “We wanted to see what other behaviors are changed with hunger, with either negative or positive consequences,” Betley told PRF. “We reasoned that individuals must prioritize the most acute threat to survival and behave accordingly,” Betley said.
So, the researchers paired two competing threats: pain and extreme hunger. Mice were deprived of food for 24 hours and then injected with formalin in the paw, a model of inflammatory pain with acute and persistent phases. Food-deprived mice displayed less paw licking in the 15 to 45 minutes following injection, whereas acute behaviors immediately after injection were similar to fed mice. The hungry mice’s responses to noxious mechanical and thermal stimuli mimicked those of fed mice, demonstrating that hunger attenuated persistent inflammatory pain but not acute pain. In addition, injection of complete Freund’s adjuvant (CFA), another model of inflammatory pain, produced hypersensitivity to thermal and mechanical stimuli in fed mice but not in food-restricted mice.
Hunger also reduced the affective component of inflammatory pain: Whereas fed mice exhibited conditioned place avoidance of formalin-paired cues, food-deprived mice did not. The hungry mice also displayed less formalin-induced immobility than fed mice.
“We didn’t expect that animals would have different responses to [acute and chronic] pain, but seeing this response specifically to longer-term pain—it made a lot of sense ethologically,” Betley said. This is because acute pain poses a high-priority threat, but an animal with persistent pain must forage and eat to avoid starvation.
Historically, researchers have looked at pain as a single entity, said Fuchs. “What’s exciting about this paper is that we are now looking at pain in a more holistic setting, as a homeostatic function. Pain is a threat to homeostasis just like hunger, so they probably activate very similar mechanisms. But how these systems interact with each other to engage goal-directed behaviors we really do not know. This study now provides an opening to understand these circuits.”
Hypothalamic neurons that link pain and hunger
To understand how hunger might dampen persistent inflammatory pain, Betley and colleagues turned to a population of hypothalamic neurons that express agouti-related peptide (AgRP), whose activation causes sated animals to eat, whereas inhibiting the neurons reduces eating in hungry mice.
“We started with the AgRP neurons because there is 20 years of work showing that their activity increases when animals are hungry, and that they are necessary and sufficient to regulate feeding. So they were the most likely suspects as an entry point to aspects of hunger that relate to pain,” Betley said.
So first author Amber Alhadeff used optogenetics to investigate the AgRP neurons. When she photostimulated normally fed mice expressing channelrhodopsin-2 (ChR2) specifically in AgRP neurons, pain behaviors were dramatically reduced during the inflammatory phase after formalin injection, as was hypersensitivity after CFA injection, compared to mice without ChR2. Acute pain behaviors did not differ from controls. Photostimulation of ChR2-expressing AgRP neurons in mice during ongoing inflammatory pain led to reduced paw licking within minutes, suggesting AgRP neuron activation rapidly directed behavior.
Conversely, inhibiting AgRP neurons by using designer receptors exclusively activated by designer drugs (DREADD) technology in food-restricted mice reduced hunger’s analgesic effect. Together, the results indicated that AgRP neurons were necessary and sufficient to inhibit inflammatory pain.
PBN is the sole target for analgesia
The researchers then wanted to identify the downstream target of AgRP neurons that produced analgesia. In a previous study, Betley and colleagues (Betley et al., 2013) had shown that the axons of AgRP neurons project to their downstream targets with one-to-one architecture, with each neuron projecting to only one brain region.
In the current study, Alhadeff painstakingly photostimulated individual subpopulations of ChR2-expressing AgRP neurons that projected to the bed nucleus of the striatum (BNST), paraventricular thalamic nucleus (PVT), paraventricular hypothalamic nucleus (PVH), or the lateral hypothalamus (LH). Doing so evoked feeding but did not inhibit pain behaviors. Stimulation of AgRP neurons that projected to the periaqueductal gray (PAG) or the central nucleus of the amygdala (CeA) had no effect on feeding or pain behaviors. But activation of the 300 or so AgRP neurons projecting solely to the lateral PBN abolished animals’ pain responses to persistent inflammatory pain after formalin injection while leaving acute pain responses intact.
“They looked at all these targets to see which produced the analgesic effect. That’s a lot of work,” Sharif-Naeini told PRF. “This is pretty convincing evidence that the PBN is critical: They essentially shut off all the pain behavior when they activated terminals in the PBN as opposed to any other area.”
There was no reason to expect that one specific subpopulation alone would inhibit pain, Betley told PRF. “When we saw that just the subset projecting to the PBN regulated pain—that was exciting because it showed that the convergence of inflammatory pain and hunger is occurring at the PBN.”
The targets downstream from the PBN that inhibit inflammatory pain remain to be seen. Fuchs wonders whether the circuit might engage the hypothalamic-pituitary-adrenal (HPA) axis. “How much of the stress system is involved? We know about stress-induced analgesia. So teasing apart the different contributions will be important,” he told PRF.
In addition to AgRP, the neurons also express the neurotransmitters gamma-aminobutyric acid (GABA) and neuropeptide Y (NPY). To determine which was responsible for dampening inflammatory pain, fed mice received PBN microinjections of GABA, AgRP, or NPY a few minutes before formalin injection. NPY provided protection against inflammatory pain, whereas GABA and AgRP had no effect on pain behaviors. Delivering an antagonist of NPY Y1 receptors to the lateral PBN eliminated the analgesic effect of hunger in food-deprived mice as well as the analgesic effect of optogenetic stimulation of ChR2-expressing AgRP neurons in fed mice. This indicated that NPY signaling by the AgRP neurons at Y1 receptors links hunger and inflammatory analgesia.
Finally, the researchers showed that in contrast to persistent inflammatory pain, acute thermal pain (from a hotplate) decreased animals’ feeding behaviors—suggesting a rank-order hierarchy of threats headed up by acute pain. In vivo calcium signaling, used as a measure of neuronal activity, showed that the activity of AgRP neurons decreased in response to food presentation, as expected, but it also decreased in response to acute thermal pain. The results indicate that AgRP neurons are bidirectionally controlled by hunger and pain signals to direct behavior.
Fasting is not a feasible way to curb pain in people, though, Betley said. After 24 hours without food, “a mouse loses 8 to 10 percent of its body weight, so that’s like five to six days without food for us. That’s massive hunger.” Instead, he said, once the endogenous pathway controlling hunger-induced analgesia is more fully characterized, it might be targeted independently of hunger, perhaps directly at the PBN or at a downstream target.
Most exciting, Betley said, “was the fact that we were able to dissociate responses to acute pain from chronic, inflammatory pain. And we demonstrated that the brain has the ability to [selectively inhibit inflammatory pain] without any drugs, so it’s up to our ingenuity to figure out how that happens and recapitulate it.”
The brain’s danger detector
The second study, led by Palmiter, focused on neurons in the lateral PBN that express calcitonin gene-related peptide (CGRP), which the group had previously shown are important drivers of appetite control. Several years ago, the group performed experiments in which they ablated hypothalamic AgRP neurons, which caused mice to starve (Luquet et al., 2005; Carter et al., 2013).
“We discovered that AgRP neurons normally inhibited CGRP-expressing PBN neurons, but in their absence, PBN neurons became hyperactive and drove the starvation phenotype,” Palmiter said. “We knew that, when active, PBN neurons inhibited feeding behavior, but they were also activated by footshock. That led us to consider their role beyond feeding.”
So, Palmiter said, “the question became, is this one homogenous population of neurons, or is it a collection of different neurons that all happen to express CGRP, with some responding to different specific stimuli?”
To find out, first author Carlos Campos injected mice in the lateral PBN with a viral vector enabling expression of a fluorescent calcium indicator dye specifically in CGRP neurons. This allowed him to visually record activity in individual neurons. All CGRP-expressing neurons in the PBN from which he recorded showed transient calcium elevations in response to repeated tail pinches, but also to paw pinches, a warm metal rod to the lip, and injection of lipopolysaccharide (LPS), which causes visceral pain.
These findings indicated that the neurons responded to various noxious stimuli regardless of the bodily site of stimulation. The neurons also encoded stimulus intensity; they responded with increasing calcium activity to more intense electric shocks to the tail and hotter noxious thermal stimuli.
CGRP-expressing neurons in the PBN responded to non-nociceptive threats as well, including subcutaneous injection of chloroquine, which causes itch. Campos then manipulated these neurons so that they could no longer release neurotransmitters. Mice with silenced CGRP-expressing PBN neurons exhibited less scratching behavior than controls in response to chloroquine and fewer swipes attempting to remove an adhesive sticker from the neck, another itch-related behavior.
“These neurons don’t care about the modality of a stimulus,” said Greg Dussor, University of Texas at Dallas, US, who was not involved in the study. “They simply signal a potentially dangerous situation.”
The researchers next restricted food overnight and recorded neuronal activity upon food presentation. When animals are in a hungry state, CGRP-expressing PBN neurons are already inhibited by AgRP inputs, but calcium signals fell further upon mice visualizing the chow.
“We noticed that their activity changed depending on cues. They were inhibited during feeding—an appetitive signal—demonstrating bidirectional control of these neurons by sensory stimuli,” Campos said. “They seem to be activated by negative valence signals and inhibited by something positive happening.” As feeding progressed, the neurons became more active, and eating declined, demonstrating their role as satiety sensors to prevent overeating.
The CGRP-expressing PBN neurons act as a general danger sensor without revealing the nature of the threat, which Palmiter likened to a home alarm. “The alarm goes off while you’re away, and you don’t know if it’s a broken window, an intruder, or a fire—you just know that something bad has happened.”
The actual and the possible
Interestingly, even the possibility of a threat activated these neurons. Although the cells were inhibited by the positive cue of familiar chow pellets, presentation of a novel, high-fat pellet enhanced calcium activity during the first exposure to the pellets.
“Even a palatable novel food—something that ultimately might become your favorite food—is scary at first,” Palmiter told PRF. “Novelty sends a warning: Taste it gingerly; don’t poison yourself.”
With subsequent exposures to the tasty food, however, the CGRP-expressing PBN neurons ceased to activate, instead becoming inhibited just as during normal feeding. Presentation of a novel inedible object—a marble—also initially triggered neuronal activity, which ebbed away with continued exposure. The marble, however, did not lead to neuronal inhibition like food did. “So these neurons get activated not only by real scary things, but also by things that are potentially scary,” Palmiter said.
The ability of CGRP-expressing PBN neurons to adapt based on repeated exposure to the novel cues showed they were sensitive to positive learning, so the researchers tested whether they were also sensitive to negative associations. Mice were conditioned to create a fear memory in a paradigm pairing a footshock with a neutral tone. The next day, the tone produced freezing—a fear behavior—along with activation of the CGRP-expressing PBN neurons, as did being placed in the shock chamber. These findings indicated the cells respond not only to a primary noxious experience but also to recall of the event. In mice with silenced CGRP-expressing PBN neurons, freezing behavior in response to the conditioned tone was rapidly extinguished.
Food for thought
Dussor described the PBN’s emerging role in pain. “The somatosensory pain pathway tells us, Where is the stimulus coming from, and what kind of pain is it? But parallel to that pathway, activity in the PBN and the amygdala seems to encode the emotional response to pain.
“It’s interesting to think about what’s actually activating these CGRP-expressing neurons, what’s upstream,” Dussor said. “Sensory inputs from the spinal cord feed into the circuit, but the fact that you can activate them with just a potentially threatening stimulus—upstream circuits from the cortex have to process the fact that the animal has never seen this before.”
Dussor also wondered how the CGRP-expressing PBN neurons might respond to chronic pain.
“If you looked at these neurons, say, a month after nerve injury, are they continuing to drive the emotional states in chronic pain? And if you silenced them to get rid of the emotional response to pain, does that take away your protective fear response to danger? You would need to selectively turn off their response to pain but not to other stimuli, and this paper suggests you can’t. Therapeutically, it’s challenging to think about. From a neuroscience perspective, it’s a very interesting observation.”
A key question remains: Are the PBN neurons described by the two studies the same or different populations? Palmiter told PRF that the CGRP neurons may well be involved, but they probably are not the primary targets of the AgRP neurons that mediate the analgesic effect.
“Our initial explorations suggest that it will likely be different neuron populations,” Betley told PRF, but they have not yet investigated the question specifically.
One way to find out, Sharif-Naeini said, would be to look in PBN neurons for CGRP and the Y1 NPY receptor to see if they are expressed in overlapping or distinct sets of cells.
Sharif-Naeini also told PRF that new technology such as optogenetics “has opened the door to map these circuits linking supraspinal processing centers—this is just the beginning. It’s an exciting time to be in neuroscience.”
Stephani Sutherland, PhD, is a neuroscientist, yogi, and freelance writer in Southern California.
Image credit: wetzkaz/123RF Stock Photo
Comments
Reza Sharif-Naeini, McGill University
How an organism maintains its
How an organism maintains its internal stability despite variations in energy, temperature, or hydration was at the core of studies by 19th-century French physiologist Claude Bernard. One of his important assertions was that “all the vital mechanisms, however varied they might be, always have one purpose, that of maintaining the integrity of the conditions of life within the internal environment.” This integrity is aggressively defended by the organism’s series of compensatory systems, which are never at rest but rather in a continuous state of high or low activity. In other words, they would be right in the middle of the ascending phase of a sigmoid curve, in which minute changes (increase or decrease) in one of the organism’s constants (hydration, temperature, circulating sodium, metabolism) would result in significant changes in physiological, affective, or behavioral responses. The non-resting state of these compensatory mechanisms led the American physiologist Walter B. Cannon to coin the term “homeostasis” (as opposed to “homeostatic”), because these systems are anything but “static.”
Two new studies not only advance our understanding of some of these compensatory mechanisms, but also shed light on how they can interact with one another to appropriately respond to changes in the internal milieu.
The study by Campos and colleagues follows a long series of outstanding research from the Palmiter group in their attempts to understand the physiological roles of subsets of neurons in the parabrachial nucleus (PBN). The work of Alhadeff and colleagues from the Betley lab, on the other hand, highlights how compensatory systems can interact and “prioritize” the stimulus that needs immediate response.
Simplified diagram of the modulation of hunger, satiety, and threat signals by the parabrachial nucleus (PBN). Excitatory and inhibitory inputs are shown in green and red, respectively. CGRP-containing PBN neurons (CGRPPBN) signal meal termination through excitatory inputs to the central amygdala (CeA). GABAergic neurons in the CeA are involved in meal termination (for a detailed description of this complex nucleus, see work from the Tonegawa group; Kim et al., 2017). Hunger signals activate AgRP neurons in the arcuate nucleus, which inhibit CGRPPBN neurons through GABA release, and shut down the anorexia pathway. After meal termination, satiety signals activate vagal afferents, which activate glutamatergic neurons in the nucleus tractus solitarius (NTS). These NTS neurons activate CGRPPBN neurons to reactivate the anorexic pathway. Acute pain signals (52°C heat in this case), perceived as more important threats than hunger, reduce the activity of AgRP neurons, thus disinhibiting PBN neurons (CGRP+ or -) to suppress appetite pathways. Other threat signals, such as painful pinch, heat, itch, or footshock, can also stimulate the food suppression pathway by activating CGRPPBN neurons (whether they do so by direct activation of CGRPPBN neurons or by inhibiting AgRP neurons remains to be determined). The PBN neurons (CGRP+ or -) are also involved in signaling inflammatory pain: In food-restricted mice, hunger becomes a more urgent threat, and NPY signaling from AgRP neurons blocks the activity of PBN neurons to reduce inflammatory pain. Figure and caption provided courtesy of Reza Sharif-Naeini.
The Campos study examines the role of a subset of neurons (CGRP-expressing) in the external lateral PBN. These neurons, termed CGRPPBN, act as a potential point of convergence of multiple sensory modalities. Their previous work had shown that these neurons are part of a complex circuit controlling appetite (see figure). The CGRPPBN neurons signal meal termination and are under the inhibition of GABAergic inputs from AgRP (agouti-related peptide) neurons in the arcuate nucleus. The latter inhibitory neurons are activated by nutritional signals related to hunger. When animals are sated, satiety-related factors activate vagal afferents that excite glutamatergic neurons in the nucleus tractus solitarius (NTS). These excitatory neurons send their terminals on to the CGRPPBN neurons (Campos et al., 2016). The anorexigenic signal from CGRPPBN neurons (glutamate and CGRP) targets CGRP receptor-expressing neurons of the central amygdala (CeA; Campos et al., 2016). Interestingly, this CGRPPBN-CeA circuit is not only used for the control of appetite but also by threat signals (Han et al., 2015). Because multiple signals used the same circuit, the question arose as to whether these different signals used segregated circuits transiting through different subsets of CGRPPBN neurons, or whether the same CGRPPBN neurons could transfer different signals.
The researchers' hypothesis was that if these signals could fit as “danger signals” coming from either cutaneous pain or food consumption above physiological capacity, then they could be encoded by the same neurons. Using in vivo visualization of calcium dynamics in awake mice, they demonstrate that the same neurons respond to feeding and various noxious stimuli, highlighting CGRPPBN neurons as a point of convergence of these aversive signals.
This convergence of aversive inputs was demonstrated in several contexts. They observed that painful thermal stimulation of the face activated the same CGRPPBN neurons that responded to tail pinch or visceral pain. Furthermore, most CGRPPBN neurons were activated by the pruritogen chloroquine. Silencing these neurons reduced chloroquine-induced scratching bouts, indicating that non-painful yet aversive/unpleasant stimuli can activate this pathway.
The role of CGRPPBN neurons in feeding was confirmed in these live calcium imaging experiments, during which calcium activity in CGRPPBN neurons of food-deprived mice was rapidly decreased when food was presented, and further decreased before each bite. After 30 minutes of feeding, CGRPPBN neurons became active again, which was temporally correlated to a decrease in food intake.
The study from Alhadeff and collaborators is a tour de force that helps us understand how two essential compensatory systems, those controlling feeding and the responses to inflammatory pain, can interact with one another. In their study, they demonstrate that activity in hunger-activated AgRP neurons of the arcuate nucleus can prevent behavioral responses to inflammatory pain via inhibition of PBN neurons through the release of neuropeptide Y (NPY). This circuit did not block acute pain, since the latter would likely be prioritized as the most salient threat. On the contrary, acute thermal pain inhibited the activity of AgRP neurons. Thus, hunger inhibits inflammatory pain, and acute pain inhibits hunger, elegantly demonstrating the prioritization of survival needs during which the most urgent threat is dealt with first.
In mice deprived of food for 24 hours, the inflammatory phase of the formalin test was significantly reduced while acute pain, or the behavioral response to the hotplate or von Frey tests, was unaffected. The interpretation is that acute pain is not inhibited by food deprivation. However, further behavioral characterization might allow one to distinguish the reflexive (withdrawal) and affective (escape behavior) components of their response to acute pain (as in Han et al., 2015). This is even more important because CGRPPBN neurons have been shown to be involved in transmitting the affective part of acute pain.
The reduction of inflammatory pain caused by food restriction was also observed in a chronic inflammatory model (the CFA model), in which both mechanical and thermal sensitization are present.
The authors subsequently examined how hunger influences the affective component of pain via a conditioned place aversion paradigm. Their results demonstrate that conditioned place aversion can be abolished if animals are food restricted. They next show that once the targets of the effect of hunger are identified, one can directly stimulate these neurons without needing to food restrict the mice. Using channelrhodopsin-2 (ChR2) expressed in the arcuate nucleus, the researchers demonstrate that photostimulation of AgRP neurons reduces formalin-induced inflammatory pain. This effect has a delay of a few minutes, suggesting its signaling may not be through activation of GABA-A receptors in the PBN. Initiating AgRP activity during a persistent inflammatory pain episode inhibits paw licking in mice. On the other hand, silencing AgRP neurons during food restriction reduces the protective effect of hunger on inflammatory pain.
Because several brain areas are innervated by AgRP neurons, the authors went through a systematic examination of each nucleus to determine which one is responsible for the decrease in inflammatory pain. These included the bed nucleus of the stria terminalis, the paraventricular thalamic nucleus, the paraventricular hypothalamic nucleus, the lateral hypothalamus, the central nucleus of the amygdala, the periaqueductal grey, and the PBN. They identify the PBN as the functional target of AgRP neurons, and demonstrate that the main transmitter essential for these effects is neuropeptide Y (NPY), acting on NPY Y1 receptor-containing neurons of the PBN. Interestingly, these AgRP neurons co-release GABA and AgRP, with the former acting as a permissive factor for feeding. Yet the suppression of inflammatory pain was independent of GABA or AgRP, indicating that two transmitters released from the same neuron can produce distinct behavioral effects.
Importantly, whether the NPY Y1 receptor-expressing neurons in the PBN are the same as the CGRP-containing neurons, which have been shown by the Palmiter group to transmit the affective component of pain, remains to be determined. If this was the case, it would provide evidence that the threat-sensing CGRPPBN neurons identified by the Palmiter group are the same neurons that inhibit hunger state (Betley group) and signal the affective component of acute pain.
In conclusion, both studies push the limits of our understanding of compensatory mechanisms and demonstrate that they can interact/compete with each other to generate an appropriate response to a threat, whether it is starvation or pain.