Social influences can profoundly modulate pain in both humans and rodents. A handful of studies have shown that when two animals experience the same pain stimulus while housed together, exposure to one another causes each animal to have a greater pain response than if they were housed alone. Now, new research shows that not only can social factors enhance pain in the setting of injury or noxious stimulation, but they can also cause pain in control mice—and without the need for the animals to ever interact.
Using three mouse models of hyperalgesia (complete Freund’s adjuvant (CFA) injection, morphine withdrawal, and alcohol withdrawal), a team of researchers led by Andrey Ryabinin, Oregon Health and Science University School of Medicine, Portland, US, demonstrates that the induction of pain in one set of mice also caused pain in a separate set of control mice—controls that were housed in the same room, in different cages, and were not exposed to a painful stimulus. This effect could be measured with mechanical, thermal, and chemical pain behavioral assays. The investigators further show that olfactory cues were responsible for the transfer of alcohol withdrawal-induced hyperalgesia.
“It’s a very interesting finding—and puzzling,” says Jeffrey Mogil, McGill University, Montreal, Canada, who was not involved in the study. “In previous studies, there had to be a pain stimulus. Here, the controls are having apparent pain hypersensitivity without ever having the stimulus in the first place, which is surprising,” continued Mogil.
The new study was published online October 19 in Science Advances.
That social interactions can modulate the experience of pain is not a new idea (Goubert et al., 2005). This has been thought of as “emotional contagion,” a form of empathy, where pain-related behaviors in one rodent directly enhance similar pain-related behaviors in another rodent when they interact with one another (Langford et al., 2006; Martin et al., 2015).
In earlier research, rats genetically predisposed to have lower pain phenotypes experienced enhanced pain behaviors following nerve injury if housed in the same cage as rats that were predisposed to have higher pain phenotypes (Raber and Devor, 2002). Similarly, mice living in the same cage had enhanced writhing behavior during the acetic acid test if they could see one another in pain (Langford et al., 2006). In a more recent investigation, naive mice housed with cagemates who had undergone sciatic nerve lesion also experienced greater pain behaviors during the acetic acid test, which the authors hypothesized was due to the stress of living with a conspecific in pain (Baptista-de-Souza et al., 2015). It’s clear, then, that social influences can modulate pain behavior, but it was unknown whether those influences also apply to animals not subjected to pain and living apart (so-called “bystander” animals), and if so, whether visual or auditory cues were necessary for that effect.
Controls feel pain, too
Ryabinin and colleagues initially set out to explore whether alcohol withdrawal caused pain in mice, as is seen in humans. In an experimental group, mice had free access to ethanol for six days, followed by a 24-hour withdrawal period without alcohol. Control mice were housed without alcohol in the same room, but in different cages. Both groups were then tested for mechanical sensitivity with Von Frey hairs (VFHs).
“We found that alcohol withdrawal indeed led to sensitivity in the experimental mice, but surprisingly, it also did in the controls,” Ryabinin said, but it was unclear why.
Monique Smith, the first author of the study and now at Stanford University, US, discussed the findings with Mary Heinricher, a pain researcher and study co-author at Oregon Health and Science University School of Medicine. “During my consultations with Mary, she agreed that it was possible for the animals to be communicating their pain, although we were all very surprised and somewhat skeptical,” said Smith.
To address this, Smith introduced a second no-alcohol control group, but one that was housed in an entirely separate room away from the first two groups. These animals continued to be pain-free.
The control animals in the same room as the experimental animals not only showed increased mechanical hyperalgesia, but also elevated chemical and thermal hyperalgesia, as measured with the formalin and hot water tail immersion tests, respectively. Similar results were also seen using CFA and morphine withdrawal models. Somehow, housing the animals that were in pain in the same room but different cages as the controls made the controls act as if they, too, were in pain.
A role for olfactory cues
Since the animals were housed individually and couldn’t see each other, the researchers wondered what other types of sensory modalities might account for the transfer of pain, hypothesizing that olfactory cues might play a role. To test this idea, Smith scooped up a small amount of dirty bedding from the alcohol withdrawal mice and placed it in an empty cage next to the cage of control mice that were in a separate room, the only group that previously did not experience pain. Even without having direct contact with this dirty bedding, these mice developed mechanical and chemical hypersensitivity within 24 hours, as measured with VFHs and the formalin test, respectively.
Smith wondered if this transfer of pain might be due to stress, but found no differences between any of the groups in plasma cortisol levels, nor in behavior on the elevated plus maze, two measures of stress. “There is something in the bedding, but we really don’t know what it is,” Ryabinin said.
What next?
The study provides further evidence that the social aspect of the biopsychosocial model of health affects pain, even in rodents. Basic scientists using animals may need to consider how this will influence their studies.
“The big message here to researchers using animals is to reconsider your housing conditions,” said Ryabinin.
Realistically, however, the practicality of housing animals in entirely separate rooms may be challenging. “[Doing so may] be even worse because there might be differences in the stress or noise conditions between the rooms,” Mogil said.
The results may prove valuable for researchers interested in using rodents to study the more common scenario seen in humans, where someone in chronic pain may live with others who are not. “People living with chronic pain patients are definitely influenced by them. Now we have a model where we can test that social influence without the requirement of an injury in both groups and try to use the findings to better understand what may happen in humans,” said Ryabinin.
Given the olfactory component, one may go so far as to say that pain could be considered contagious. “This paper suggests that it might be true, but obviously we want to treat this suggestion with a lot of healthy skepticism. For one thing, humans are not nearly as olfactory as mice,” Mogil explained.
Image credit: anyaivanova/123RF Stock Photo
Correction: The original sentence descrbing the results from Raber and Devor included the phrase “that also underwent nerve injury” at the end of the sentence. In the corrected version, that phrase has been deleted, in order to clarify that the effect of housing with animals predisposed to have higher pain phenotypes did not rely on these animals undergoing surgery. That is, housing low-pain phenotype animals that underwent surgery with high-pain phenotype animals that did not undergo surgery was sufficient to enhance the pain state of the low-pain phenotype animals.
Comments
Jeffrey Mogil, McGill University
So, the big question I have
So, the big question I have is this. If this were true, and pervasive, shouldn't we all be seeing mechanical allodynia in our control groups? Forget about housing in the same room, in all my studies mice in the control group are in the same cage as mice in the CFA (or whatever other injury) group. Why are we not all seeing this routinely? Anyone have large relevant datasets they can use to address this?
Andrey Ryabinin, Oregon Health and Science University
We are sorry for just now
This comment is co-authored by Andrey Ryabinin and Mary Heinricher, Oregon Health & Science University.
We are sorry for just now noticing this comment by Jeff Mogil. We have replicated this effect in over a dozen independent experiments, using multiple persistent pain paradigms, and more than one researcher in the lab has replicated the basic observation. Further, the Mogil group has itself noted that simply returning an animal to a cage following nociceptive testing results in shorter response latencies in animals subsequently tested from that same cage (Chesler et al., 2002). Although this observation was not pursued in any systematic way, it is surprising that Dr. Mogil did not take his own prior findings into account when assessing our work.
Turning Dr. Mogil’s comment on itself, one might ask why the influence of experimenter sex reported in his own studies has not been seen more widely?
One answer might be in a recent suggestion that neuronal circuits normally engaged by physiologically and behaviorally relevant variables (e.g., social structure, availability of food, predation pressure) might be “hijacked” in random ways by artificial laboratory variables such as experimenter sex, ultrasonic noise, and other environmental features (Lahvis, 2017). Studies focused on more biologically relevant social stimuli are therefore more likely to be more meaningful.
We invite replication by others and hope that other researchers who have approached us after our publication and/or presentation of the data saying that they have hints of similar effects in their studies will speak up.
Robert Sorge, University of Alabama at Birmingham
I admit that I read the paper
I admit that I read the paper by Ryabinin and colleagues with interest, given our own work with olfactory cues. With the publication of their paper, I looked into our own data where we (like in Dr. Mogil's example) house our "injured" animals in the same cages as the control animals. We do not show this effect or even a hint of this aside from a mild effect on the first day that is likely due to stress. To echo Jeff's question, this seems like something people would have noticed at this point with thousands of mice being tested every day.
To comment on the authors' question as to why the experimenter sex effect was not noticed, I think they are comparing apples to oranges here. The experimenter sex effect was present, but not the focus of the studies in which it may have had an effect. The sensitivity of the control group is a directly assessed variable in all studies that utilize a control group and, thus, under direct scrutiny.
In this case, the comparison is invalid. The question raised by Jeff is that this effect is something that should have been recognized earlier since it is a variable being measured. The fact that it has not been seen, but seems to be so clearly evident in the paper by Ryabinin and colleagues, suggests that other factors may be at work. In the end, I don’t know what to think of it or where the effect is coming from. The authors show a very robust social effect, but I am unconvinced that their assertion is the whole story or we should have been seeing this for years. At the very least, this has opened the door to exploring more factors that may be unexplored up to this point.
Andrey Ryabinin, Oregon Health and Science University
We humbly disagree that the
This comment is co-authored by Andrey Ryabinin and Mary Heinricher, Oregon Health & Science University.
We humbly disagree that the previous absence of detected effects of the sex of the experimenter (identified by Sorge and Mogil) and of the effect of social transfer (identified in my lab) represent incomparable "apples and oranges." In both cases, the effects were for the first time identified when the appropriate factors were specifically addressed by experimental design. In Dr. Sorge's paper, the experiments were designed to use either male or female handlers. In our studies, the experiments included an additional "Separate" control group. This control group was housed in a separate room from experimental animals. The surprising matching of thresholds between experimental animals and "bystander controls" housed in the same room was observed only in the alcohol-withdrawal experiments, in which hyperalgesia is relatively mild. In the more robust inflammatory model, which is certainly more standard in the field, the matching between experimental animals and "bystander controls" was far from complete. This phenomenon is not therefore a quantal effect.
The critical point is that it is impossible to identify the effects of social influence of experimental animals on bystander controls in the absence of a separately-housed control group. Dr. Sorge argues that social transfer of hyperalgesia was not observed in his experiments, but those experiments do not involve controls sitting in a separate room. We suggest that it is unknown how Dr. Sorge’s "co-housed" controls would have behaved if they had been housed in a separate room.
In any case, we are glad that our findings are receiving attention in the field, and raising awareness of potential social influences on pain measures in rodents.